Merge branch 'ls/mailinfo'
authorJunio C Hamano <gitster@pobox.com>
Thu, 17 Jul 2008 00:49:53 +0000 (17:49 -0700)
committerJunio C Hamano <gitster@pobox.com>
Thu, 17 Jul 2008 00:49:53 +0000 (17:49 -0700)
* ls/mailinfo:
  git-mailinfo: use strbuf's instead of fixed buffers
  Add some useful functions for strbuf manipulation.
  Make some strbuf_*() struct strbuf arguments const.

Conflicts:
builtin-mailinfo.c

builtin-mailinfo.c
strbuf.c
strbuf.h

index a4f99c8..4b82615 100644 (file)
@@ -5,14 +5,15 @@
 #include "cache.h"
 #include "builtin.h"
 #include "utf8.h"
+#include "strbuf.h"
 
 static FILE *cmitmsg, *patchfile, *fin, *fout;
 
 static int keep_subject;
 static const char *metainfo_charset;
-static char line[1000];
-static char name[1000];
-static char email[1000];
+static struct strbuf line = STRBUF_INIT;
+static struct strbuf name = STRBUF_INIT;
+static struct strbuf email = STRBUF_INIT;
 
 static enum  {
        TE_DONTCARE, TE_QP, TE_BASE64,
@@ -21,74 +22,79 @@ static enum  {
        TYPE_TEXT, TYPE_OTHER,
 } message_type;
 
-static char charset[256];
+static struct strbuf charset = STRBUF_INIT;
 static int patch_lines;
-static char **p_hdr_data, **s_hdr_data;
+static struct strbuf **p_hdr_data, **s_hdr_data;
 
 #define MAX_HDR_PARSED 10
 #define MAX_BOUNDARIES 5
 
-static char *sanity_check(char *name, char *email)
+static void get_sane_name(struct strbuf *out, struct strbuf *name, struct strbuf *email)
 {
-       int len = strlen(name);
-       if (len < 3 || len > 60)
-               return email;
-       if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
-               return email;
-       return name;
+       struct strbuf *src = name;
+       if (name->len < 3 || 60 < name->len || strchr(name->buf, '@') ||
+               strchr(name->buf, '<') || strchr(name->buf, '>'))
+               src = email;
+       else if (name == out)
+               return;
+       strbuf_reset(out);
+       strbuf_addbuf(out, src);
 }
 
-static int bogus_from(char *line)
+static void parse_bogus_from(const struct strbuf *line)
 {
        /* John Doe <johndoe> */
-       char *bra, *ket, *dst, *cp;
 
+       char *bra, *ket;
        /* This is fallback, so do not bother if we already have an
         * e-mail address.
         */
-       if (*email)
-               return 0;
+       if (email.len)
+               return;
 
-       bra = strchr(line, '<');
+       bra = strchr(line->buf, '<');
        if (!bra)
-               return 0;
+               return;
        ket = strchr(bra, '>');
        if (!ket)
-               return 0;
+               return;
 
-       for (dst = email, cp = bra+1; cp < ket; )
-               *dst++ = *cp++;
-       *dst = 0;
-       for (cp = line; isspace(*cp); cp++)
-               ;
-       for (bra--; isspace(*bra); bra--)
-               *bra = 0;
-       cp = sanity_check(cp, email);
-       strcpy(name, cp);
-       return 1;
+       strbuf_reset(&email);
+       strbuf_add(&email, bra + 1, ket - bra - 1);
+
+       strbuf_reset(&name);
+       strbuf_add(&name, line->buf, bra - line->buf);
+       strbuf_trim(&name);
+       get_sane_name(&name, &name, &email);
 }
 
-static int handle_from(char *in_line)
+static void handle_from(const struct strbuf *from)
 {
-       char line[1000];
        char *at;
-       char *dst;
+       size_t el;
+       struct strbuf f;
 
-       strcpy(line, in_line);
-       at = strchr(line, '@');
-       if (!at)
-               return bogus_from(line);
+       strbuf_init(&f, from->len);
+       strbuf_addbuf(&f, from);
+
+       at = strchr(f.buf, '@');
+       if (!at) {
+               parse_bogus_from(from);
+               return;
+       }
 
        /*
         * If we already have one email, don't take any confusing lines
         */
-       if (*email && strchr(at+1, '@'))
-               return 0;
+       if (email.len && strchr(at + 1, '@')) {
+               strbuf_release(&f);
+               return;
+       }
 
        /* Pick up the string around '@', possibly delimited with <>
-        * pair; that is the email part.  White them out while copying.
+        * pair; that is the email part.
         */
-       while (at > line) {
+       while (at > f.buf) {
                char c = at[-1];
                if (isspace(c))
                        break;
@@ -98,56 +104,35 @@ static int handle_from(char *in_line)
                }
                at--;
        }
-       dst = email;
-       for (;;) {
-               unsigned char c = *at;
-               if (!c || c == '>' || isspace(c)) {
-                       if (c == '>')
-                               *at = ' ';
-                       break;
-               }
-               *at++ = ' ';
-               *dst++ = c;
-       }
-       *dst++ = 0;
+       el = strcspn(at, " \n\t\r\v\f>");
+       strbuf_reset(&email);
+       strbuf_add(&email, at, el);
+       strbuf_remove(&f, at - f.buf, el + 1);
 
        /* The remainder is name.  It could be "John Doe <john.doe@xz>"
-        * or "john.doe@xz (John Doe)", but we have whited out the
+        * or "john.doe@xz (John Doe)", but we have removed the
         * email part, so trim from both ends, possibly removing
         * the () pair at the end.
         */
-       at = line + strlen(line);
-       while (at > line) {
-               unsigned char c = *--at;
-               if (!isspace(c)) {
-                       at[(c == ')') ? 0 : 1] = 0;
-                       break;
-               }
-       }
+       strbuf_trim(&f);
+       if (f.buf[0] == '(')
+               strbuf_remove(&name, 0, 1);
+       if (f.len && f.buf[f.len - 1] == ')')
+               strbuf_setlen(&f, f.len - 1);
 
-       at = line;
-       for (;;) {
-               unsigned char c = *at;
-               if (!c || !isspace(c)) {
-                       if (c == '(')
-                               at++;
-                       break;
-               }
-               at++;
-       }
-       at = sanity_check(at, email);
-       strcpy(name, at);
-       return 1;
+       get_sane_name(&name, &f, &email);
+       strbuf_release(&f);
 }
 
-static int handle_header(char *line, char *data, int ofs)
+static void handle_header(struct strbuf **out, const struct strbuf *line)
 {
-       if (!line || !data)
-               return 1;
+       if (!*out) {
+               *out = xmalloc(sizeof(struct strbuf));
+               strbuf_init(*out, line->len);
+       } else
+               strbuf_reset(*out);
 
-       strcpy(data, line+ofs);
-
-       return 0;
+       strbuf_addbuf(*out, line);
 }
 
 /* NOTE NOTE NOTE.  We do not claim we do full MIME.  We just attempt
@@ -156,13 +141,13 @@ static int handle_header(char *line, char *data, int ofs)
  * case insensitively.
  */
 
-static int slurp_attr(const char *line, const char *name, char *attr)
+static int slurp_attr(const char *line, const char *name, struct strbuf *attr)
 {
        const char *ends, *ap = strcasestr(line, name);
        size_t sz;
 
        if (!ap) {
-               *attr = 0;
+               strbuf_setlen(attr, 0);
                return 0;
        }
        ap += strlen(name);
@@ -173,182 +158,172 @@ static int slurp_attr(const char *line, const char *name, char *attr)
        else
                ends = "; \t";
        sz = strcspn(ap, ends);
-       memcpy(attr, ap, sz);
-       attr[sz] = 0;
+       strbuf_add(attr, ap, sz);
        return 1;
 }
 
-struct content_type {
-       char *boundary;
-       int boundary_len;
-};
-
-static struct content_type content[MAX_BOUNDARIES];
+static struct strbuf *content[MAX_BOUNDARIES];
 
-static struct content_type *content_top = content;
+static struct strbuf **content_top = content;
 
-static int handle_content_type(char *line)
+static void handle_content_type(struct strbuf *line)
 {
-       char boundary[256];
+       struct strbuf *boundary = xmalloc(sizeof(struct strbuf));
+       strbuf_init(boundary, line->len);
 
-       if (strcasestr(line, "text/") == NULL)
+       if (!strcasestr(line->buf, "text/"))
                 message_type = TYPE_OTHER;
-       if (slurp_attr(line, "boundary=", boundary + 2)) {
-               memcpy(boundary, "--", 2);
+       if (slurp_attr(line->buf, "boundary=", boundary)) {
+               strbuf_insert(boundary, 0, "--", 2);
                if (content_top++ >= &content[MAX_BOUNDARIES]) {
                        fprintf(stderr, "Too many boundaries to handle\n");
                        exit(1);
                }
-               content_top->boundary_len = strlen(boundary);
-               content_top->boundary = xmalloc(content_top->boundary_len+1);
-               strcpy(content_top->boundary, boundary);
+               *content_top = boundary;
+               boundary = NULL;
        }
-       if (slurp_attr(line, "charset=", charset)) {
-               int i, c;
-               for (i = 0; (c = charset[i]) != 0; i++)
-                       charset[i] = tolower(c);
+       if (slurp_attr(line->buf, "charset=", &charset))
+               strbuf_tolower(&charset);
+
+       if (boundary) {
+               strbuf_release(boundary);
+               free(boundary);
        }
-       return 0;
 }
 
-static int handle_content_transfer_encoding(char *line)
+static void handle_content_transfer_encoding(const struct strbuf *line)
 {
-       if (strcasestr(line, "base64"))
+       if (strcasestr(line->buf, "base64"))
                transfer_encoding = TE_BASE64;
-       else if (strcasestr(line, "quoted-printable"))
+       else if (strcasestr(line->buf, "quoted-printable"))
                transfer_encoding = TE_QP;
        else
                transfer_encoding = TE_DONTCARE;
-       return 0;
 }
 
-static int is_multipart_boundary(const char *line)
+static int is_multipart_boundary(const struct strbuf *line)
 {
-       return (!memcmp(line, content_top->boundary, content_top->boundary_len));
+       return !strbuf_cmp(line, *content_top);
 }
 
-static int eatspace(char *line)
+static void cleanup_subject(struct strbuf *subject)
 {
-       int len = strlen(line);
-       while (len > 0 && isspace(line[len-1]))
-               line[--len] = 0;
-       return len;
-}
-
-static char *cleanup_subject(char *subject)
-{
-       for (;;) {
-               char *p;
-               int len, remove;
-               switch (*subject) {
+       char *pos;
+       size_t remove;
+       while (subject->len) {
+               switch (*subject->buf) {
                case 'r': case 'R':
-                       if (!memcmp("e:", subject+1, 2)) {
-                               subject += 3;
+                       if (subject->len <= 3)
+                               break;
+                       if (!memcmp(subject->buf + 1, "e:", 2)) {
+                               strbuf_remove(subject, 0, 3);
                                continue;
                        }
                        break;
                case ' ': case '\t': case ':':
-                       subject++;
+                       strbuf_remove(subject, 0, 1);
                        continue;
-
                case '[':
-                       p = strchr(subject, ']');
-                       if (!p) {
-                               subject++;
-                               continue;
-                       }
-                       len = strlen(p);
-                       remove = p - subject;
-                       if (remove <= len *2) {
-                               subject = p+1;
-                               continue;
-                       }
+                       if ((pos = strchr(subject->buf, ']'))) {
+                               remove = pos - subject->buf + 1;
+                               /* Don't remove too much. */
+                               if (remove <= (subject->len - remove + 1) * 2) {
+                                       strbuf_remove(subject, 0, remove);
+                                       continue;
+                               }
+                       } else
+                               strbuf_remove(subject, 0, 1);
                        break;
                }
-               eatspace(subject);
-               return subject;
+               strbuf_trim(subject);
+               return;
        }
 }
 
-static void cleanup_space(char *buf)
+static void cleanup_space(struct strbuf *sb)
 {
-       unsigned char c;
-       while ((c = *buf) != 0) {
-               buf++;
-               if (isspace(c)) {
-                       buf[-1] = ' ';
-                       c = *buf;
-                       while (isspace(c)) {
-                               int len = strlen(buf);
-                               memmove(buf, buf+1, len);
-                               c = *buf;
-                       }
+       size_t pos, cnt;
+       for (pos = 0; pos < sb->len; pos++) {
+               if (isspace(sb->buf[pos])) {
+                       sb->buf[pos] = ' ';
+                       for (cnt = 0; isspace(sb->buf[pos + cnt + 1]); cnt++);
+                       strbuf_remove(sb, pos + 1, cnt);
                }
        }
 }
 
-static void decode_header(char *it, unsigned itsize);
+static void decode_header(struct strbuf *line);
 static const char *header[MAX_HDR_PARSED] = {
        "From","Subject","Date",
 };
 
-static int check_header(char *line, unsigned linesize, char **hdr_data, int overwrite)
+static inline int cmp_header(const struct strbuf *line, const char *hdr)
 {
-       int i;
+       int len = strlen(hdr);
+       return !strncasecmp(line->buf, hdr, len) && line->len > len &&
+                       line->buf[len] == ':' && isspace(line->buf[len + 1]);
+}
 
+static int check_header(const struct strbuf *line,
+                               struct strbuf *hdr_data[], int overwrite)
+{
+       int i, ret = 0, len;
+       struct strbuf sb = STRBUF_INIT;
        /* search for the interesting parts */
        for (i = 0; header[i]; i++) {
                int len = strlen(header[i]);
-               if ((!hdr_data[i] || overwrite) &&
-                   !strncasecmp(line, header[i], len) &&
-                   line[len] == ':' && isspace(line[len + 1])) {
+               if ((!hdr_data[i] || overwrite) && cmp_header(line, header[i])) {
                        /* Unwrap inline B and Q encoding, and optionally
                         * normalize the meta information to utf8.
                         */
-                       decode_header(line + len + 2, linesize - len - 2);
-                       hdr_data[i] = xmalloc(1000 * sizeof(char));
-                       if (! handle_header(line, hdr_data[i], len + 2)) {
-                               return 1;
-                       }
+                       strbuf_add(&sb, line->buf + len + 2, line->len - len - 2);
+                       decode_header(&sb);
+                       handle_header(&hdr_data[i], &sb);
+                       ret = 1;
+                       goto check_header_out;
                }
        }
 
        /* Content stuff */
-       if (!strncasecmp(line, "Content-Type", 12) &&
-               line[12] == ':' && isspace(line[12 + 1])) {
-               decode_header(line + 12 + 2, linesize - 12 - 2);
-               if (! handle_content_type(line)) {
-                       return 1;
-               }
-       }
-       if (!strncasecmp(line, "Content-Transfer-Encoding", 25) &&
-               line[25] == ':' && isspace(line[25 + 1])) {
-               decode_header(line + 25 + 2, linesize - 25 - 2);
-               if (! handle_content_transfer_encoding(line)) {
-                       return 1;
-               }
+       if (cmp_header(line, "Content-Type")) {
+               len = strlen("Content-Type: ");
+               strbuf_add(&sb, line->buf + len, line->len - len);
+               decode_header(&sb);
+               strbuf_insert(&sb, 0, "Content-Type: ", len);
+               handle_content_type(&sb);
+               ret = 1;
+               goto check_header_out;
+       }
+       if (cmp_header(line, "Content-Transfer-Encoding")) {
+               len = strlen("Content-Transfer-Encoding: ");
+               strbuf_add(&sb, line->buf + len, line->len - len);
+               decode_header(&sb);
+               handle_content_transfer_encoding(&sb);
+               ret = 1;
+               goto check_header_out;
        }
 
        /* for inbody stuff */
-       if (!memcmp(">From", line, 5) && isspace(line[5]))
-               return 1;
-       if (!memcmp("[PATCH]", line, 7) && isspace(line[7])) {
+       if (!prefixcmp(line->buf, ">From") && isspace(line->buf[5])) {
+               ret = 1; /* Should this return 0? */
+               goto check_header_out;
+       }
+       if (!prefixcmp(line->buf, "[PATCH]") && isspace(line->buf[7])) {
                for (i = 0; header[i]; i++) {
                        if (!memcmp("Subject", header[i], 7)) {
-                               if (!hdr_data[i])
-                                       hdr_data[i] = xmalloc(linesize + 20);
-                               if (! handle_header(line, hdr_data[i], 0)) {
-                                       return 1;
-                               }
+                               handle_header(&hdr_data[i], line);
+                               ret = 1;
+                               goto check_header_out;
                        }
                }
        }
 
-       /* no match */
-       return 0;
+check_header_out:
+       strbuf_release(&sb);
+       return ret;
 }
 
-static int is_rfc2822_header(char *line)
+static int is_rfc2822_header(const struct strbuf *line)
 {
        /*
         * The section that defines the loosest possible
@@ -359,15 +334,15 @@ static int is_rfc2822_header(char *line)
         * ftext = %d33-57 / %59-126
         */
        int ch;
-       char *cp = line;
+       char *cp = line->buf;
 
        /* Count mbox From headers as headers */
-       if (!memcmp(line, "From ", 5) || !memcmp(line, ">From ", 6))
+       if (!prefixcmp(cp, "From ") || !prefixcmp(cp, ">From "))
                return 1;
 
        while ((ch = *cp++)) {
                if (ch == ':')
-                       return cp != line;
+                       return 1;
                if ((33 <= ch && ch <= 57) ||
                    (59 <= ch && ch <= 126))
                        continue;
@@ -376,34 +351,20 @@ static int is_rfc2822_header(char *line)
        return 0;
 }
 
-/*
- * sz is size of 'line' buffer in bytes.  Must be reasonably
- * long enough to hold one physical real-world e-mail line.
- */
-static int read_one_header_line(char *line, int sz, FILE *in)
+static int read_one_header_line(struct strbuf *line, FILE *in)
 {
-       int len;
-
-       /*
-        * We will read at most (sz-1) bytes and then potentially
-        * re-add NUL after it.  Accessing line[sz] after this is safe
-        * and we can allow len to grow up to and including sz.
-        */
-       sz--;
-
        /* Get the first part of the line. */
-       if (!fgets(line, sz, in))
+       if (strbuf_getline(line, in, '\n'))
                return 0;
 
        /*
         * Is it an empty line or not a valid rfc2822 header?
         * If so, stop here, and return false ("not a header")
         */
-       len = eatspace(line);
-       if (!len || !is_rfc2822_header(line)) {
+       strbuf_rtrim(line);
+       if (!line->len || !is_rfc2822_header(line)) {
                /* Re-add the newline */
-               line[len] = '\n';
-               line[len + 1] = '\0';
+               strbuf_addch(line, '\n');
                return 0;
        }
 
@@ -412,65 +373,53 @@ static int read_one_header_line(char *line, int sz, FILE *in)
         * Yuck, 2822 header "folding"
         */
        for (;;) {
-               int peek, addlen;
-               static char continuation[1000];
+               int peek;
+               struct strbuf continuation = STRBUF_INIT;
 
                peek = fgetc(in); ungetc(peek, in);
                if (peek != ' ' && peek != '\t')
                        break;
-               if (!fgets(continuation, sizeof(continuation), in))
+               if (strbuf_getline(&continuation, in, '\n'))
                        break;
-               addlen = eatspace(continuation);
-               if (len < sz - 1) {
-                       if (addlen >= sz - len)
-                               addlen = sz - len - 1;
-                       memcpy(line + len, continuation, addlen);
-                       line[len] = '\n';
-                       len += addlen;
-               }
+               continuation.buf[0] = '\n';
+               strbuf_rtrim(&continuation);
+               strbuf_addbuf(line, &continuation);
        }
-       line[len] = 0;
 
        return 1;
 }
 
-static int decode_q_segment(char *in, char *ot, unsigned otsize, char *ep, int rfc2047)
+static struct strbuf *decode_q_segment(const struct strbuf *q_seg, int rfc2047)
 {
-       char *otbegin = ot;
-       char *otend = ot + otsize;
+       const char *in = q_seg->buf;
        int c;
-       while ((c = *in++) != 0 && (in <= ep)) {
-               if (ot == otend) {
-                       *--ot = '\0';
-                       return -1;
-               }
+       struct strbuf *out = xmalloc(sizeof(struct strbuf));
+       strbuf_init(out, q_seg->len);
+
+       while ((c = *in++) != 0) {
                if (c == '=') {
                        int d = *in++;
                        if (d == '\n' || !d)
                                break; /* drop trailing newline */
-                       *ot++ = ((hexval(d) << 4) | hexval(*in++));
+                       strbuf_addch(out, (hexval(d) << 4) | hexval(*in++));
                        continue;
                }
                if (rfc2047 && c == '_') /* rfc2047 4.2 (2) */
                        c = 0x20;
-               *ot++ = c;
+               strbuf_addch(out, c);
        }
-       *ot = 0;
-       return (ot - otbegin);
+       return out;
 }
 
-static int decode_b_segment(char *in, char *ot, unsigned otsize, char *ep)
+static struct strbuf *decode_b_segment(const struct strbuf *b_seg)
 {
        /* Decode in..ep, possibly in-place to ot */
        int c, pos = 0, acc = 0;
-       char *otbegin = ot;
-       char *otend = ot + otsize;
+       const char *in = b_seg->buf;
+       struct strbuf *out = xmalloc(sizeof(struct strbuf));
+       strbuf_init(out, b_seg->len);
 
-       while ((c = *in++) != 0 && (in <= ep)) {
-               if (ot == otend) {
-                       *--ot = '\0';
-                       return -1;
-               }
+       while ((c = *in++) != 0) {
                if (c == '+')
                        c = 62;
                else if (c == '/')
@@ -495,21 +444,20 @@ static int decode_b_segment(char *in, char *ot, unsigned otsize, char *ep)
                        acc = (c << 2);
                        break;
                case 1:
-                       *ot++ = (acc | (c >> 4));
+                       strbuf_addch(out, (acc | (c >> 4)));
                        acc = (c & 15) << 4;
                        break;
                case 2:
-                       *ot++ = (acc | (c >> 2));
+                       strbuf_addch(out, (acc | (c >> 2)));
                        acc = (c & 3) << 6;
                        break;
                case 3:
-                       *ot++ = (acc | c);
+                       strbuf_addch(out, (acc | c));
                        acc = pos = 0;
                        break;
                }
        }
-       *ot = 0;
-       return (ot - otbegin);
+       return out;
 }
 
 /*
@@ -523,16 +471,16 @@ static int decode_b_segment(char *in, char *ot, unsigned otsize, char *ep)
  * Otherwise, we default to assuming it is Latin1 for historical
  * reasons.
  */
-static const char *guess_charset(const char *line, const char *target_charset)
+static const char *guess_charset(const struct strbuf *line, const char *target_charset)
 {
        if (is_encoding_utf8(target_charset)) {
-               if (is_utf8(line))
+               if (is_utf8(line->buf))
                        return NULL;
        }
        return "latin1";
 }
 
-static void convert_to_utf8(char *line, unsigned linesize, const char *charset)
+static void convert_to_utf8(struct strbuf *line, const char *charset)
 {
        char *out;
 
@@ -544,112 +492,118 @@ static void convert_to_utf8(char *line, unsigned linesize, const char *charset)
 
        if (!strcmp(metainfo_charset, charset))
                return;
-       out = reencode_string(line, metainfo_charset, charset);
+       out = reencode_string(line->buf, metainfo_charset, charset);
        if (!out)
                die("cannot convert from %s to %s\n",
                    charset, metainfo_charset);
-       strlcpy(line, out, linesize);
-       free(out);
+       strbuf_attach(line, out, strlen(out), strlen(out));
 }
 
-static int decode_header_bq(char *it, unsigned itsize)
+static int decode_header_bq(struct strbuf *it)
 {
-       char *in, *out, *ep, *cp, *sp;
-       char outbuf[1000];
+       char *in, *ep, *cp;
+       struct strbuf outbuf = STRBUF_INIT, *dec;
+       struct strbuf charset_q = STRBUF_INIT, piecebuf = STRBUF_INIT;
        int rfc2047 = 0;
 
-       in = it;
-       out = outbuf;
-       while ((ep = strstr(in, "=?")) != NULL) {
-               int sz, encoding;
-               char charset_q[256], piecebuf[256];
+       in = it->buf;
+       while (in - it->buf <= it->len && (ep = strstr(in, "=?")) != NULL) {
+               int encoding;
+               strbuf_reset(&charset_q);
+               strbuf_reset(&piecebuf);
                rfc2047 = 1;
 
                if (in != ep) {
-                       sz = ep - in;
-                       memcpy(out, in, sz);
-                       out += sz;
-                       in += sz;
+                       strbuf_add(&outbuf, in, ep - in);
+                       in = ep;
                }
                /* E.g.
                 * ep : "=?iso-2022-jp?B?GyR...?= foo"
                 * ep : "=?ISO-8859-1?Q?Foo=FCbar?= baz"
                 */
                ep += 2;
-               cp = strchr(ep, '?');
-               if (!cp)
-                       return rfc2047; /* no munging */
-               for (sp = ep; sp < cp; sp++)
-                       charset_q[sp - ep] = tolower(*sp);
-               charset_q[cp - ep] = 0;
+
+               if (ep - it->buf >= it->len || !(cp = strchr(ep, '?')))
+                       goto decode_header_bq_out;
+
+               if (cp + 3 - it->buf > it->len)
+                       goto decode_header_bq_out;
+               strbuf_add(&charset_q, ep, cp - ep);
+               strbuf_tolower(&charset_q);
+
                encoding = cp[1];
                if (!encoding || cp[2] != '?')
-                       return rfc2047; /* no munging */
+                       goto decode_header_bq_out;
                ep = strstr(cp + 3, "?=");
                if (!ep)
-                       return rfc2047; /* no munging */
+                       goto decode_header_bq_out;
+               strbuf_add(&piecebuf, cp + 3, ep - cp - 3);
                switch (tolower(encoding)) {
                default:
-                       return rfc2047; /* no munging */
+                       goto decode_header_bq_out;
                case 'b':
-                       sz = decode_b_segment(cp + 3, piecebuf, sizeof(piecebuf), ep);
+                       dec = decode_b_segment(&piecebuf);
                        break;
                case 'q':
-                       sz = decode_q_segment(cp + 3, piecebuf, sizeof(piecebuf), ep, 1);
+                       dec = decode_q_segment(&piecebuf, 1);
                        break;
                }
-               if (sz < 0)
-                       return rfc2047;
                if (metainfo_charset)
-                       convert_to_utf8(piecebuf, sizeof(piecebuf), charset_q);
+                       convert_to_utf8(dec, charset_q.buf);
 
-               sz = strlen(piecebuf);
-               if (outbuf + sizeof(outbuf) <= out + sz)
-                       return rfc2047; /* no munging */
-               strcpy(out, piecebuf);
-               out += sz;
+               strbuf_addbuf(&outbuf, dec);
+               strbuf_release(dec);
+               free(dec);
                in = ep + 2;
        }
-       strcpy(out, in);
-       strlcpy(it, outbuf, itsize);
+       strbuf_addstr(&outbuf, in);
+       strbuf_reset(it);
+       strbuf_addbuf(it, &outbuf);
+decode_header_bq_out:
+       strbuf_release(&outbuf);
+       strbuf_release(&charset_q);
+       strbuf_release(&piecebuf);
        return rfc2047;
 }
 
-static void decode_header(char *it, unsigned itsize)
+static void decode_header(struct strbuf *it)
 {
-
-       if (decode_header_bq(it, itsize))
+       if (decode_header_bq(it))
                return;
        /* otherwise "it" is a straight copy of the input.
         * This can be binary guck but there is no charset specified.
         */
        if (metainfo_charset)
-               convert_to_utf8(it, itsize, "");
+               convert_to_utf8(it, "");
 }
 
-static int decode_transfer_encoding(char *line, unsigned linesize, int inputlen)
+static void decode_transfer_encoding(struct strbuf *line)
 {
-       char *ep;
+       struct strbuf *ret;
 
        switch (transfer_encoding) {
        case TE_QP:
-               ep = line + inputlen;
-               return decode_q_segment(line, line, linesize, ep, 0);
+               ret = decode_q_segment(line, 0);
+               break;
        case TE_BASE64:
-               ep = line + inputlen;
-               return decode_b_segment(line, line, linesize, ep);
+               ret = decode_b_segment(line);
+               break;
        case TE_DONTCARE:
        default:
-               return inputlen;
+               return;
        }
+       strbuf_reset(line);
+       strbuf_addbuf(line, ret);
+       strbuf_release(ret);
+       free(ret);
 }
 
-static int handle_filter(char *line, unsigned linesize, int linelen);
+static void handle_filter(struct strbuf *line);
 
 static int find_boundary(void)
 {
-       while(fgets(line, sizeof(line), fin) != NULL) {
-               if (is_multipart_boundary(line))
+       while (!strbuf_getline(&line, fin, '\n')) {
+               if (is_multipart_boundary(&line))
                        return 1;
        }
        return 0;
@@ -657,12 +611,17 @@ static int find_boundary(void)
 
 static int handle_boundary(void)
 {
-       char newline[]="\n";
+       struct strbuf newline = STRBUF_INIT;
+
+       strbuf_addch(&newline, '\n');
 again:
-       if (!memcmp(line+content_top->boundary_len, "--", 2)) {
+       if (line.len >= (*content_top)->len + 2 &&
+           !memcmp(line.buf + (*content_top)->len, "--", 2)) {
                /* we hit an end boundary */
                /* pop the current boundary off the stack */
-               free(content_top->boundary);
+               strbuf_release(*content_top);
+               free(*content_top);
+               *content_top = NULL;
 
                /* technically won't happen as is_multipart_boundary()
                   will fail first.  But just in case..
@@ -672,7 +631,8 @@ again:
                                        "can't recover\n");
                        exit(1);
                }
-               handle_filter(newline, sizeof(newline), strlen(newline));
+               handle_filter(&newline);
+               strbuf_release(&newline);
 
                /* skip to the next boundary */
                if (!find_boundary())
@@ -682,39 +642,44 @@ again:
 
        /* set some defaults */
        transfer_encoding = TE_DONTCARE;
-       charset[0] = 0;
+       strbuf_reset(&charset);
        message_type = TYPE_TEXT;
 
        /* slurp in this section's info */
-       while (read_one_header_line(line, sizeof(line), fin))
-               check_header(line, sizeof(line), p_hdr_data, 0);
+       while (read_one_header_line(&line, fin))
+               check_header(&line, p_hdr_data, 0);
 
+       strbuf_release(&newline);
        /* eat the blank line after section info */
-       return (fgets(line, sizeof(line), fin) != NULL);
+       return (strbuf_getline(&line, fin, '\n') == 0);
 }
 
-static inline int patchbreak(const char *line)
+static inline int patchbreak(const struct strbuf *line)
 {
+       size_t i;
+
        /* Beginning of a "diff -" header? */
-       if (!memcmp("diff -", line, 6))
+       if (!prefixcmp(line->buf, "diff -"))
                return 1;
 
        /* CVS "Index: " line? */
-       if (!memcmp("Index: ", line, 7))
+       if (!prefixcmp(line->buf, "Index: "))
                return 1;
 
        /*
         * "--- <filename>" starts patches without headers
         * "---<sp>*" is a manual separator
         */
-       if (!memcmp("---", line, 3)) {
-               line += 3;
+       if (line->len < 4)
+               return 0;
+
+       if (!prefixcmp(line->buf, "---")) {
                /* space followed by a filename? */
-               if (line[0] == ' ' && !isspace(line[1]))
+               if (line->buf[3] == ' ' && !isspace(line->buf[4]))
                        return 1;
                /* Just whitespace? */
-               for (;;) {
-                       unsigned char c = *line++;
+               for (i = 3; i < line->len; i++) {
+                       unsigned char c = line->buf[i];
                        if (c == '\n')
                                return 1;
                        if (!isspace(c))
@@ -725,32 +690,24 @@ static inline int patchbreak(const char *line)
        return 0;
 }
 
-
-static int handle_commit_msg(char *line, unsigned linesize)
+static int handle_commit_msg(struct strbuf *line)
 {
        static int still_looking = 1;
-       char *endline = line + linesize;
 
        if (!cmitmsg)
                return 0;
 
        if (still_looking) {
-               char *cp = line;
-               if (isspace(*line)) {
-                       for (cp = line + 1; *cp; cp++) {
-                               if (!isspace(*cp))
-                                       break;
-                       }
-                       if (!*cp)
-                               return 0;
-               }
-               if ((still_looking = check_header(cp, endline - cp, s_hdr_data, 0)) != 0)
+               strbuf_ltrim(line);
+               if (!line->len)
+                       return 0;
+               if ((still_looking = check_header(line, s_hdr_data, 0)) != 0)
                        return 0;
        }
 
        /* normalize the log message to UTF-8. */
        if (metainfo_charset)
-               convert_to_utf8(line, endline - line, charset);
+               convert_to_utf8(line, charset.buf);
 
        if (patchbreak(line)) {
                fclose(cmitmsg);
@@ -758,142 +715,132 @@ static int handle_commit_msg(char *line, unsigned linesize)
                return 1;
        }
 
-       fputs(line, cmitmsg);
+       fputs(line->buf, cmitmsg);
        return 0;
 }
 
-static int handle_patch(char *line, int len)
+static void handle_patch(const struct strbuf *line)
 {
-       fwrite(line, 1, len, patchfile);
+       fwrite(line->buf, 1, line->len, patchfile);
        patch_lines++;
-       return 0;
 }
 
-static int handle_filter(char *line, unsigned linesize, int linelen)
+static void handle_filter(struct strbuf *line)
 {
        static int filter = 0;
 
-       /* filter tells us which part we left off on
-        * a non-zero return indicates we hit a filter point
-        */
+       /* filter tells us which part we left off on */
        switch (filter) {
        case 0:
-               if (!handle_commit_msg(line, linesize))
+               if (!handle_commit_msg(line))
                        break;
                filter++;
        case 1:
-               if (!handle_patch(line, linelen))
-                       break;
-               filter++;
-       default:
-               return 1;
+               handle_patch(line);
+               break;
        }
-
-       return 0;
 }
 
 static void handle_body(void)
 {
-       int rc = 0;
-       static char newline[2000];
-       static char *np = newline;
-       int len = strlen(line);
+       int len = 0;
+       struct strbuf prev = STRBUF_INIT;
 
        /* Skip up to the first boundary */
-       if (content_top->boundary) {
+       if (*content_top) {
                if (!find_boundary())
-                       return;
+                       goto handle_body_out;
        }
 
        do {
+               strbuf_setlen(&line, line.len + len);
+
                /* process any boundary lines */
-               if (content_top->boundary && is_multipart_boundary(line)) {
+               if (*content_top && is_multipart_boundary(&line)) {
                        /* flush any leftover */
-                       if (np != newline)
-                               handle_filter(newline, sizeof(newline),
-                                             np - newline);
+                       if (line.len)
+                               handle_filter(&line);
+
                        if (!handle_boundary())
-                               return;
-                       len = strlen(line);
+                               goto handle_body_out;
                }
 
                /* Unwrap transfer encoding */
-               len = decode_transfer_encoding(line, sizeof(line), len);
-               if (len < 0) {
-                       error("Malformed input line");
-                       return;
-               }
+               decode_transfer_encoding(&line);
 
                switch (transfer_encoding) {
                case TE_BASE64:
                case TE_QP:
                {
-                       char *op = line;
+                       struct strbuf **lines, **it, *sb;
+
+                       /* Prepend any previous partial lines */
+                       strbuf_insert(&line, 0, prev.buf, prev.len);
+                       strbuf_reset(&prev);
 
                        /* binary data most likely doesn't have newlines */
                        if (message_type != TYPE_TEXT) {
-                               rc = handle_filter(line, sizeof(line), len);
+                               handle_filter(&line);
                                break;
                        }
-
                        /*
                         * This is a decoded line that may contain
                         * multiple new lines.  Pass only one chunk
                         * at a time to handle_filter()
                         */
-                       do {
-                               while (op < line + len && *op != '\n')
-                                       *np++ = *op++;
-                               *np = *op;
-                               if (*np != 0) {
-                                       /* should be sitting on a new line */
-                                       *(++np) = 0;
-                                       op++;
-                                       rc = handle_filter(newline, sizeof(newline), np - newline);
-                                       np = newline;
-                               }
-                       } while (op < line + len);
+                       lines = strbuf_split(&line, '\n');
+                       for (it = lines; (sb = *it); it++) {
+                               if (*(it + 1) == NULL) /* The last line */
+                                       if (sb->buf[sb->len - 1] != '\n') {
+                                               /* Partial line, save it for later. */
+                                               strbuf_addbuf(&prev, sb);
+                                               break;
+                                       }
+                               handle_filter(sb);
+                       }
                        /*
-                        * The partial chunk is saved in newline and will be
+                        * The partial chunk is saved in "prev" and will be
                         * appended by the next iteration of read_line_with_nul().
                         */
+                       strbuf_list_free(lines);
                        break;
                }
                default:
-                       rc = handle_filter(line, sizeof(line), len);
+                       handle_filter(&line);
                }
-               if (rc)
-                       /* nothing left to filter */
-                       break;
-       } while ((len = read_line_with_nul(line, sizeof(line), fin)));
 
-       return;
+               strbuf_reset(&line);
+               if (strbuf_avail(&line) < 100)
+                       strbuf_grow(&line, 100);
+       } while ((len = read_line_with_nul(line.buf, strbuf_avail(&line), fin)));
+
+handle_body_out:
+       strbuf_release(&prev);
 }
 
-static void output_header_lines(FILE *fout, const char *hdr, char *data)
+static void output_header_lines(FILE *fout, const char *hdr, const struct strbuf *data)
 {
+       const char *sp = data->buf;
        while (1) {
-               char *ep = strchr(data, '\n');
+               char *ep = strchr(sp, '\n');
                int len;
                if (!ep)
-                       len = strlen(data);
+                       len = strlen(sp);
                else
-                       len = ep - data;
-               fprintf(fout, "%s: %.*s\n", hdr, len, data);
+                       len = ep - sp;
+               fprintf(fout, "%s: %.*s\n", hdr, len, sp);
                if (!ep)
                        break;
-               data = ep + 1;
+               sp = ep + 1;
        }
 }
 
 static void handle_info(void)
 {
-       char *sub;
-       char *hdr;
+       struct strbuf *hdr;
        int i;
 
        for (i = 0; header[i]; i++) {
-
                /* only print inbody headers if we output a patch file */
                if (patch_lines && s_hdr_data[i])
                        hdr = s_hdr_data[i];
@@ -903,20 +850,18 @@ static void handle_info(void)
                        continue;
 
                if (!memcmp(header[i], "Subject", 7)) {
-                       if (keep_subject)
-                               sub = hdr;
-                       else {
-                               sub = cleanup_subject(hdr);
-                               cleanup_space(sub);
+                       if (!keep_subject) {
+                               cleanup_subject(hdr);
+                               cleanup_space(hdr);
                        }
-                       output_header_lines(fout, "Subject", sub);
+                       output_header_lines(fout, "Subject", hdr);
                } else if (!memcmp(header[i], "From", 4)) {
                        handle_from(hdr);
-                       fprintf(fout, "Author: %s\n", name);
-                       fprintf(fout, "Email: %s\n", email);
+                       fprintf(fout, "Author: %s\n", name.buf);
+                       fprintf(fout, "Email: %s\n", email.buf);
                } else {
                        cleanup_space(hdr);
-                       fprintf(fout, "%s: %s\n", header[i], hdr);
+                       fprintf(fout, "%s: %s\n", header[i], hdr->buf);
                }
        }
        fprintf(fout, "\n");
@@ -943,8 +888,8 @@ static int mailinfo(FILE *in, FILE *out, int ks, const char *encoding,
                return -1;
        }
 
-       p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
-       s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(char *));
+       p_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*p_hdr_data));
+       s_hdr_data = xcalloc(MAX_HDR_PARSED, sizeof(*s_hdr_data));
 
        do {
                peek = fgetc(in);
@@ -952,8 +897,8 @@ static int mailinfo(FILE *in, FILE *out, int ks, const char *encoding,
        ungetc(peek, in);
 
        /* process the email header */
-       while (read_one_header_line(line, sizeof(line), fin))
-               check_header(line, sizeof(line), p_hdr_data, 1);
+       while (read_one_header_line(&line, fin))
+               check_header(&line, p_hdr_data, 1);
 
        handle_body();
        handle_info();
index 4aed752..720737d 100644 (file)
--- a/strbuf.c
+++ b/strbuf.c
@@ -60,6 +60,18 @@ void strbuf_grow(struct strbuf *sb, size_t extra)
        ALLOC_GROW(sb->buf, sb->len + extra + 1, sb->alloc);
 }
 
+void strbuf_trim(struct strbuf *sb)
+{
+       char *b = sb->buf;
+       while (sb->len > 0 && isspace((unsigned char)sb->buf[sb->len - 1]))
+               sb->len--;
+       while (sb->len > 0 && isspace(*b)) {
+               b++;
+               sb->len--;
+       }
+       memmove(sb->buf, b, sb->len);
+       sb->buf[sb->len] = '\0';
+}
 void strbuf_rtrim(struct strbuf *sb)
 {
        while (sb->len > 0 && isspace((unsigned char)sb->buf[sb->len - 1]))
@@ -67,7 +79,65 @@ void strbuf_rtrim(struct strbuf *sb)
        sb->buf[sb->len] = '\0';
 }
 
-int strbuf_cmp(struct strbuf *a, struct strbuf *b)
+void strbuf_ltrim(struct strbuf *sb)
+{
+       char *b = sb->buf;
+       while (sb->len > 0 && isspace(*b)) {
+               b++;
+               sb->len--;
+       }
+       memmove(sb->buf, b, sb->len);
+       sb->buf[sb->len] = '\0';
+}
+
+void strbuf_tolower(struct strbuf *sb)
+{
+       int i;
+       for (i = 0; i < sb->len; i++)
+               sb->buf[i] = tolower(sb->buf[i]);
+}
+
+struct strbuf **strbuf_split(const struct strbuf *sb, int delim)
+{
+       int alloc = 2, pos = 0;
+       char *n, *p;
+       struct strbuf **ret;
+       struct strbuf *t;
+
+       ret = xcalloc(alloc, sizeof(struct strbuf *));
+       p = n = sb->buf;
+       while (n < sb->buf + sb->len) {
+               int len;
+               n = memchr(n, delim, sb->len - (n - sb->buf));
+               if (pos + 1 >= alloc) {
+                       alloc = alloc * 2;
+                       ret = xrealloc(ret, sizeof(struct strbuf *) * alloc);
+               }
+               if (!n)
+                       n = sb->buf + sb->len - 1;
+               len = n - p + 1;
+               t = xmalloc(sizeof(struct strbuf));
+               strbuf_init(t, len);
+               strbuf_add(t, p, len);
+               ret[pos] = t;
+               ret[++pos] = NULL;
+               p = ++n;
+       }
+       return ret;
+}
+
+void strbuf_list_free(struct strbuf **sbs)
+{
+       struct strbuf **s = sbs;
+
+       while (*s) {
+               strbuf_release(*s);
+               free(*s++);
+       }
+       free(sbs);
+}
+
+int strbuf_cmp(const struct strbuf *a, const struct strbuf *b)
 {
        int cmp;
        if (a->len < b->len) {
index faec229..0c6ffad 100644 (file)
--- a/strbuf.h
+++ b/strbuf.h
@@ -61,7 +61,7 @@ static inline void strbuf_swap(struct strbuf *a, struct strbuf *b) {
 }
 
 /*----- strbuf size related -----*/
-static inline size_t strbuf_avail(struct strbuf *sb) {
+static inline size_t strbuf_avail(const struct strbuf *sb) {
        return sb->alloc ? sb->alloc - sb->len - 1 : 0;
 }
 
@@ -77,8 +77,14 @@ static inline void strbuf_setlen(struct strbuf *sb, size_t len) {
 #define strbuf_reset(sb)  strbuf_setlen(sb, 0)
 
 /*----- content related -----*/
+extern void strbuf_trim(struct strbuf *);
 extern void strbuf_rtrim(struct strbuf *);
-extern int strbuf_cmp(struct strbuf *, struct strbuf *);
+extern void strbuf_ltrim(struct strbuf *);
+extern int strbuf_cmp(const struct strbuf *, const struct strbuf *);
+extern void strbuf_tolower(struct strbuf *);
+
+extern struct strbuf **strbuf_split(const struct strbuf *, int delim);
+extern void strbuf_list_free(struct strbuf **);
 
 /*----- add data in your buffer -----*/
 static inline void strbuf_addch(struct strbuf *sb, int c) {
@@ -98,7 +104,7 @@ extern void strbuf_add(struct strbuf *, const void *, size_t);
 static inline void strbuf_addstr(struct strbuf *sb, const char *s) {
        strbuf_add(sb, s, strlen(s));
 }
-static inline void strbuf_addbuf(struct strbuf *sb, struct strbuf *sb2) {
+static inline void strbuf_addbuf(struct strbuf *sb, const struct strbuf *sb2) {
        strbuf_add(sb, sb2->buf, sb2->len);
 }
 extern void strbuf_adddup(struct strbuf *sb, size_t pos, size_t len);