hash-object: cleanup handling of command line options
[gitweb.git] / pretty.c
index 9fbd73f748c6cd250b5e9534168072a1cea88a85..997f5837d521ded5f158d12abc0c025f74e83341 100644 (file)
--- a/pretty.c
+++ b/pretty.c
@@ -282,51 +282,59 @@ static char *logmsg_reencode(const struct commit *commit,
        return out;
 }
 
-static void format_person_part(struct strbuf *sb, char part,
+static size_t format_person_part(struct strbuf *sb, char part,
                                const char *msg, int len)
 {
+       /* currently all placeholders have same length */
+       const int placeholder_len = 2;
        int start, end, tz = 0;
-       unsigned long date;
+       unsigned long date = 0;
        char *ep;
 
-       /* parse name */
+       /* advance 'end' to point to email start delimiter */
        for (end = 0; end < len && msg[end] != '<'; end++)
                ; /* do nothing */
-       start = end + 1;
-       while (end > 0 && isspace(msg[end - 1]))
-               end--;
+
+       /*
+        * When end points at the '<' that we found, it should have
+        * matching '>' later, which means 'end' must be strictly
+        * below len - 1.
+        */
+       if (end >= len - 2)
+               goto skip;
+
        if (part == 'n') {      /* name */
+               while (end > 0 && isspace(msg[end - 1]))
+                       end--;
                strbuf_add(sb, msg, end);
-               return;
+               return placeholder_len;
        }
+       start = ++end; /* save email start position */
 
-       if (start >= len)
-               return;
-
-       /* parse email */
-       for (end = start + 1; end < len && msg[end] != '>'; end++)
+       /* advance 'end' to point to email end delimiter */
+       for ( ; end < len && msg[end] != '>'; end++)
                ; /* do nothing */
 
        if (end >= len)
-               return;
+               goto skip;
 
        if (part == 'e') {      /* email */
                strbuf_add(sb, msg + start, end - start);
-               return;
+               return placeholder_len;
        }
 
-       /* parse date */
+       /* advance 'start' to point to date start delimiter */
        for (start = end + 1; start < len && isspace(msg[start]); start++)
                ; /* do nothing */
        if (start >= len)
-               return;
+               goto skip;
        date = strtoul(msg + start, &ep, 10);
        if (msg + start == ep)
-               return;
+               goto skip;
 
        if (part == 't') {      /* date, UNIX timestamp */
                strbuf_add(sb, msg + start, ep - (msg + start));
-               return;
+               return placeholder_len;
        }
 
        /* parse tz */
@@ -341,48 +349,136 @@ static void format_person_part(struct strbuf *sb, char part,
        switch (part) {
        case 'd':       /* date */
                strbuf_addstr(sb, show_date(date, tz, DATE_NORMAL));
-               return;
+               return placeholder_len;
        case 'D':       /* date, RFC2822 style */
                strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
-               return;
+               return placeholder_len;
        case 'r':       /* date, relative */
                strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
-               return;
+               return placeholder_len;
        case 'i':       /* date, ISO 8601 */
                strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
-               return;
+               return placeholder_len;
        }
+
+skip:
+       /*
+        * bogus commit, 'sb' cannot be updated, but we still need to
+        * compute a valid return value.
+        */
+       if (part == 'n' || part == 'e' || part == 't' || part == 'd'
+           || part == 'D' || part == 'r' || part == 'i')
+               return placeholder_len;
+
+       return 0; /* unknown placeholder */
 }
 
-static void format_commit_item(struct strbuf *sb, const char *placeholder,
-                               void *context)
+struct chunk {
+       size_t off;
+       size_t len;
+};
+
+struct format_commit_context {
+       const struct commit *commit;
+
+       /* These offsets are relative to the start of the commit message. */
+       int commit_header_parsed;
+       struct chunk subject;
+       struct chunk author;
+       struct chunk committer;
+       struct chunk encoding;
+       size_t body_off;
+
+       /* The following ones are relative to the result struct strbuf. */
+       struct chunk abbrev_commit_hash;
+       struct chunk abbrev_tree_hash;
+       struct chunk abbrev_parent_hashes;
+};
+
+static int add_again(struct strbuf *sb, struct chunk *chunk)
 {
-       const struct commit *commit = context;
-       struct commit_list *p;
+       if (chunk->len) {
+               strbuf_adddup(sb, chunk->off, chunk->len);
+               return 1;
+       }
+
+       /*
+        * We haven't seen this chunk before.  Our caller is surely
+        * going to add it the hard way now.  Remember the most likely
+        * start of the to-be-added chunk: the current end of the
+        * struct strbuf.
+        */
+       chunk->off = sb->len;
+       return 0;
+}
+
+static void parse_commit_header(struct format_commit_context *context)
+{
+       const char *msg = context->commit->buffer;
        int i;
        enum { HEADER, SUBJECT, BODY } state;
+
+       for (i = 0, state = HEADER; msg[i] && state < BODY; i++) {
+               int eol;
+               for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
+                       ; /* do nothing */
+
+               if (state == SUBJECT) {
+                       context->subject.off = i;
+                       context->subject.len = eol - i;
+                       i = eol;
+               }
+               if (i == eol) {
+                       state++;
+                       /* strip empty lines */
+                       while (msg[eol] == '\n' && msg[eol + 1] == '\n')
+                               eol++;
+               } else if (!prefixcmp(msg + i, "author ")) {
+                       context->author.off = i + 7;
+                       context->author.len = eol - i - 7;
+               } else if (!prefixcmp(msg + i, "committer ")) {
+                       context->committer.off = i + 10;
+                       context->committer.len = eol - i - 10;
+               } else if (!prefixcmp(msg + i, "encoding ")) {
+                       context->encoding.off = i + 9;
+                       context->encoding.len = eol - i - 9;
+               }
+               i = eol;
+               if (!msg[i])
+                       break;
+       }
+       context->body_off = i;
+       context->commit_header_parsed = 1;
+}
+
+static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
+                               void *context)
+{
+       struct format_commit_context *c = context;
+       const struct commit *commit = c->commit;
        const char *msg = commit->buffer;
+       struct commit_list *p;
 
        /* these are independent of the commit */
        switch (placeholder[0]) {
        case 'C':
-               switch (placeholder[3]) {
-               case 'd':       /* red */
+               if (!prefixcmp(placeholder + 1, "red")) {
                        strbuf_addstr(sb, "\033[31m");
-                       return;
-               case 'e':       /* green */
+                       return 4;
+               } else if (!prefixcmp(placeholder + 1, "green")) {
                        strbuf_addstr(sb, "\033[32m");
-                       return;
-               case 'u':       /* blue */
+                       return 6;
+               } else if (!prefixcmp(placeholder + 1, "blue")) {
                        strbuf_addstr(sb, "\033[34m");
-                       return;
-               case 's':       /* reset color */
+                       return 5;
+               } else if (!prefixcmp(placeholder + 1, "reset")) {
                        strbuf_addstr(sb, "\033[m");
-                       return;
-               }
+                       return 6;
+               } else
+                       return 0;
        case 'n':               /* newline */
                strbuf_addch(sb, '\n');
-               return;
+               return 1;
        }
 
        /* these depend on the commit */
@@ -392,120 +488,84 @@ static void format_commit_item(struct strbuf *sb, const char *placeholder,
        switch (placeholder[0]) {
        case 'H':               /* commit hash */
                strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
-               return;
+               return 1;
        case 'h':               /* abbreviated commit hash */
+               if (add_again(sb, &c->abbrev_commit_hash))
+                       return 1;
                strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
                                                     DEFAULT_ABBREV));
-               return;
+               c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
+               return 1;
        case 'T':               /* tree hash */
                strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
-               return;
+               return 1;
        case 't':               /* abbreviated tree hash */
+               if (add_again(sb, &c->abbrev_tree_hash))
+                       return 1;
                strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
                                                     DEFAULT_ABBREV));
-               return;
+               c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
+               return 1;
        case 'P':               /* parent hashes */
                for (p = commit->parents; p; p = p->next) {
                        if (p != commit->parents)
                                strbuf_addch(sb, ' ');
                        strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
                }
-               return;
+               return 1;
        case 'p':               /* abbreviated parent hashes */
+               if (add_again(sb, &c->abbrev_parent_hashes))
+                       return 1;
                for (p = commit->parents; p; p = p->next) {
                        if (p != commit->parents)
                                strbuf_addch(sb, ' ');
                        strbuf_addstr(sb, find_unique_abbrev(
                                        p->item->object.sha1, DEFAULT_ABBREV));
                }
-               return;
+               c->abbrev_parent_hashes.len = sb->len -
+                                             c->abbrev_parent_hashes.off;
+               return 1;
        case 'm':               /* left/right/bottom */
                strbuf_addch(sb, (commit->object.flags & BOUNDARY)
                                 ? '-'
                                 : (commit->object.flags & SYMMETRIC_LEFT)
                                 ? '<'
                                 : '>');
-               return;
+               return 1;
        }
 
        /* For the rest we have to parse the commit header. */
-       for (i = 0, state = HEADER; msg[i] && state < BODY; i++) {
-               int eol;
-               for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
-                       ; /* do nothing */
+       if (!c->commit_header_parsed)
+               parse_commit_header(c);
 
-               if (state == SUBJECT) {
-                       if (placeholder[0] == 's') {
-                               strbuf_add(sb, msg + i, eol - i);
-                               return;
-                       }
-                       i = eol;
-               }
-               if (i == eol) {
-                       state++;
-                       /* strip empty lines */
-                       while (msg[eol + 1] == '\n')
-                               eol++;
-               } else if (!prefixcmp(msg + i, "author ")) {
-                       if (placeholder[0] == 'a') {
-                               format_person_part(sb, placeholder[1],
-                                                  msg + i + 7, eol - i - 7);
-                               return;
-                       }
-               } else if (!prefixcmp(msg + i, "committer ")) {
-                       if (placeholder[0] == 'c') {
-                               format_person_part(sb, placeholder[1],
-                                                  msg + i + 10, eol - i - 10);
-                               return;
-                       }
-               } else if (!prefixcmp(msg + i, "encoding ")) {
-                       if (placeholder[0] == 'e') {
-                               strbuf_add(sb, msg + i + 9, eol - i - 9);
-                               return;
-                       }
-               }
-               i = eol;
+       switch (placeholder[0]) {
+       case 's':       /* subject */
+               strbuf_add(sb, msg + c->subject.off, c->subject.len);
+               return 1;
+       case 'a':       /* author ... */
+               return format_person_part(sb, placeholder[1],
+                                  msg + c->author.off, c->author.len);
+       case 'c':       /* committer ... */
+               return format_person_part(sb, placeholder[1],
+                                  msg + c->committer.off, c->committer.len);
+       case 'e':       /* encoding */
+               strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
+               return 1;
+       case 'b':       /* body */
+               strbuf_addstr(sb, msg + c->body_off);
+               return 1;
        }
-       if (msg[i] && placeholder[0] == 'b')    /* body */
-               strbuf_addstr(sb, msg + i);
+       return 0;       /* unknown placeholder */
 }
 
 void format_commit_message(const struct commit *commit,
                            const void *format, struct strbuf *sb)
 {
-       const char *placeholders[] = {
-               "H",            /* commit hash */
-               "h",            /* abbreviated commit hash */
-               "T",            /* tree hash */
-               "t",            /* abbreviated tree hash */
-               "P",            /* parent hashes */
-               "p",            /* abbreviated parent hashes */
-               "an",           /* author name */
-               "ae",           /* author email */
-               "ad",           /* author date */
-               "aD",           /* author date, RFC2822 style */
-               "ar",           /* author date, relative */
-               "at",           /* author date, UNIX timestamp */
-               "ai",           /* author date, ISO 8601 */
-               "cn",           /* committer name */
-               "ce",           /* committer email */
-               "cd",           /* committer date */
-               "cD",           /* committer date, RFC2822 style */
-               "cr",           /* committer date, relative */
-               "ct",           /* committer date, UNIX timestamp */
-               "ci",           /* committer date, ISO 8601 */
-               "e",            /* encoding */
-               "s",            /* subject */
-               "b",            /* body */
-               "Cred",         /* red */
-               "Cgreen",       /* green */
-               "Cblue",        /* blue */
-               "Creset",       /* reset color */
-               "n",            /* newline */
-               "m",            /* left/right/bottom */
-               NULL
-       };
-       strbuf_expand(sb, format, placeholders, format_commit_item, (void *)commit);
+       struct format_commit_context context;
+
+       memset(&context, 0, sizeof(context));
+       context.commit = commit;
+       strbuf_expand(sb, format, format_commit_item, &context);
 }
 
 static void pp_header(enum cmit_fmt fmt,