1 file changed, 144 insertions, 134 deletions
diff --git a/source/parsing.c b/source/parsing.c
index 0d63b51..de2798a 100644
--- a/source/parsing.c
+++ b/source/parsing.c
@@ -1,127 +1,61 @@
-/* parsing.c: parsing of config files
- *
- * Copyright (C) 2006-2014 cgit Development Team <cgit@lists.zx2c4.com>
- *
- * Licensed under GNU General Public License v2
- *   (see LICENSE.txt for full license text)
+/*
+ * The two kinds of raw text cgit reads before it can render a page, the path
+ * of the incoming request and the body of a commit or a tag object. Reading
+ * the request path settles which repository and which page the request works
+ * on. Commit and tag objects arrive as header lines, a blank line, and then
+ * the message, so the readers here walk the headers, copy out the fields the
+ * pages need, and re-encode the message into the page encoding.
  */
 
 #define USE_THE_REPOSITORY_VARIABLE
 
 #include "cgit.h"
+#include "parsing.h"
+#include "shared.h"
 
-/*
- * url syntax: [repo ['/' cmd [ '/' path]]]
- *   repo: any valid repo url, may contain '/'
- *   cmd:  log | commit | diff | tree | view | blob | snapshot
- *   path: any valid path, may contain '/'
- *
- */
-void cgit_parse_url(const char *url)
-{
-	char *c, *cmd, *p, *buf;
-	struct cgit_repo *repo;
-
-	if (!url || url[0] == '\0')
-		return;
-
-	ctx.qry.page = NULL;
-	ctx.repo = cgit_get_repoinfo(url);
-	if (ctx.repo) {
-		ctx.qry.repo = ctx.repo->url;
-		return;
-	}
-
-	buf = xstrdup(url);
-	cmd = NULL;
-	c = strchr(buf, '/');
-	while (c) {
-		c[0] = '\0';
-		repo = cgit_get_repoinfo(buf);
-		if (repo) {
-			ctx.repo = repo;
-			cmd = c;
-		}
-		c[0] = '/';
-		c = strchr(c + 1, '/');
-	}
-
-	if (ctx.repo) {
-		ctx.qry.repo = ctx.repo->url;
-		p = strchr(cmd + 1, '/');
-		if (p) {
-			p[0] = '\0';
-			if (p[1])
-				ctx.qry.path = cgit_trim_end(p + 1, '/');
-		}
-		if (cmd[1])
-			ctx.qry.page = xstrdup(cmd + 1);
-	}
-	free(buf);
-}
-
-static char *substr(const char *head, const char *tail)
+static char *substr(const char *start, const char *end)
 {
 	size_t len;
 	char *buf;
 
-	if (tail < head)
+	if (end < start)
 		return xstrdup("");
-	// head points into the commit buffer, so strlcpy would measure the
-	// whole remaining commit just to copy a name or a subject off the front
-	// of it.
-	len = tail - head;
+	// start points into the object buffer, so strlcpy would measure the
+	// whole rest of the object to copy a name off the front of it.
+	len = end - start;
 	buf = xmalloc(len + 1);
-	memcpy(buf, head, len);
+	memcpy(buf, start, len);
 	buf[len] = '\0';
 	return buf;
 }
 
-static void parse_user(const char *t, char **name, char **email, unsigned long *date, int *tz)
+static void parse_user(const char *line, char **name, char **email,
+		       timestamp_t *date, int *tz)
 {
 	struct ident_split ident;
-	unsigned email_len;
+	struct strbuf address = STRBUF_INIT;
+	ptrdiff_t email_len;
 
-	if (!split_ident_line(&ident, t, strchrnul(t, '\n') - t)) {
+	if (!split_ident_line(&ident, line, strchrnul(line, '\n') - line)) {
 		*name = substr(ident.name_begin, ident.name_end);
 
+		// Assembled rather than formatted. The length is a pointer
+		// difference, and the precision of a %.*s conversion has to be
+		// an int, which cannot hold one on a 64 bit host.
 		email_len = ident.mail_end - ident.mail_begin;
-		*email = xmalloc(strlen("<") + email_len + strlen(">") + 1);
-		xsnprintf(*email, email_len + 3, "<%.*s>", email_len, ident.mail_begin);
+		strbuf_addch(&address, '<');
+		if (email_len > 0)
+			strbuf_add(&address, ident.mail_begin, email_len);
+		strbuf_addch(&address, '>');
+		*email = strbuf_detach(&address, NULL);
 
 		if (ident.date_begin)
-			*date = strtoul(ident.date_begin, NULL, 10);
+			*date = parse_timestamp(ident.date_begin, NULL, 10);
 		if (ident.tz_begin)
 			*tz = atoi(ident.tz_begin);
 	}
 }
 
-#ifdef NO_ICONV
-#define reencode(a, b, c)
-#else
-static const char *reencode(char **txt, const char *src_enc, const char *dst_enc)
-{
-	char *tmp;
-
-	if (!txt)
-		return NULL;
-
-	if (!*txt || !src_enc || !dst_enc)
-		return *txt;
-
-	/* no encoding needed if src_enc equals dst_enc */
-	if (!strcasecmp(src_enc, dst_enc))
-		return *txt;
-
-	tmp = reencode_string(*txt, dst_enc, src_enc);
-	if (tmp) {
-		free(*txt);
-		*txt = tmp;
-	}
-	return *txt;
-}
-#endif
-
 static const char *next_header_line(const char *p)
 {
 	p = strchr(p, '\n');
@@ -135,17 +69,89 @@ static int end_of_header(const char *p)
 	return !p || (*p == '\n');
 }
 
+/*
+ * A git built without iconv still offers reencode_string, where it always
+ * fails, so the field is left in whatever encoding it arrived in.
+ */
+static const char *reencode(char **text, const char *from, const char *to)
+{
+	char *converted;
+
+	if (!text)
+		return NULL;
+
+	if (!*text || !from || !to)
+		return *text;
+
+	if (!strcasecmp(from, to))
+		return *text;
+
+	converted = reencode_string(*text, to, from);
+	if (converted) {
+		free(*text);
+		*text = converted;
+	}
+	return *text;
+}
+
+/*
+ * A repository url may itself contain slashes, so every slash-separated
+ * prefix is looked up and the longest one that names a repository wins.
+ */
+void cgit_parse_url(const char *url)
+{
+	char *buf, *slash, *repo_end, *page_end;
+	struct cgit_repo *repo;
+
+	if (!url || url[0] == '\0')
+		return;
+
+	ctx.qry.page = NULL;
+	ctx.repo = cgit_get_repoinfo(url);
+	if (ctx.repo) {
+		ctx.qry.repo = ctx.repo->url;
+		return;
+	}
+
+	buf = xstrdup(url);
+	repo_end = NULL;
+	slash = strchr(buf, '/');
+	while (slash) {
+		slash[0] = '\0';
+		repo = cgit_get_repoinfo(buf);
+		if (repo) {
+			ctx.repo = repo;
+			repo_end = slash;
+		}
+		slash[0] = '/';
+		slash = strchr(slash + 1, '/');
+	}
+
+	if (ctx.repo) {
+		ctx.qry.repo = ctx.repo->url;
+		page_end = strchr(repo_end + 1, '/');
+		if (page_end) {
+			page_end[0] = '\0';
+			if (page_end[1])
+				ctx.qry.path = cgit_trim_end(page_end + 1, '/');
+		}
+		if (repo_end[1])
+			ctx.qry.page = xstrdup(repo_end + 1);
+	}
+	free(buf);
+}
+
 struct commitinfo *cgit_parse_commit(struct commit *commit)
 {
-	struct commitinfo *ret;
+	struct commitinfo *info;
 	const char *p = repo_get_commit_buffer(the_repository, commit, NULL);
-	const char *t;
+	const char *eol;
 
-	ret = xcalloc(1, sizeof(struct commitinfo));
-	ret->commit = commit;
+	info = xcalloc(1, sizeof(struct commitinfo));
+	info->commit = commit;
 
 	if (!p)
-		return ret;
+		return info;
 
 	if (!skip_prefix(p, "tree ", &p))
 		die("Bad commit: %s", oid_to_hex(&commit->object.oid));
@@ -155,27 +161,29 @@ struct commitinfo *cgit_parse_commit(struct commit *commit)
 		p += the_hash_algo->hexsz + 1;
 
 	if (p && skip_prefix(p, "author ", &p)) {
-		parse_user(p, &ret->author, &ret->author_email,
-			&ret->author_date, &ret->author_tz);
+		parse_user(p, &info->author, &info->author_email,
+			&info->author_date, &info->author_tz);
 		p = next_header_line(p);
 	}
 
 	if (p && skip_prefix(p, "committer ", &p)) {
-		parse_user(p, &ret->committer, &ret->committer_email,
-			&ret->committer_date, &ret->committer_tz);
+		parse_user(p, &info->committer, &info->committer_email,
+			&info->committer_date, &info->committer_tz);
 		p = next_header_line(p);
 	}
 
 	if (p && skip_prefix(p, "encoding ", &p)) {
-		t = strchr(p, '\n');
-		if (t) {
-			ret->msg_encoding = substr(p, t + 1);
-			p = t + 1;
+		eol = strchr(p, '\n');
+		if (eol) {
+			info->msg_encoding = substr(p, eol + 1);
+			p = eol + 1;
 		}
 	}
 
-	if (!ret->msg_encoding)
-		ret->msg_encoding = xstrdup("UTF-8");
+	// Git only writes the header when the message is in something other
+	// than UTF-8, so its absence means UTF-8.
+	if (!info->msg_encoding)
+		info->msg_encoding = xstrdup("UTF-8");
 
 	while (!end_of_header(p))
 		p = next_header_line(p);
@@ -183,27 +191,28 @@ struct commitinfo *cgit_parse_commit(struct commit *commit)
 		p++;
 
 	if (p) {
-		t = strchrnul(p, '\n');
-		ret->subject = substr(p, t);
-		while (*t == '\n')
-			t++;
-		ret->msg = xstrdup(t);
+		eol = strchrnul(p, '\n');
+		info->subject = substr(p, eol);
+		while (*eol == '\n')
+			eol++;
+		info->msg = xstrdup(eol);
 	} else {
-		// A crafted commit can end right after its headers with no
-		// message. Keep subject and msg as empty strings so callers
-		// can treat them as text unconditionally.
-		ret->subject = xstrdup("");
-		ret->msg = xstrdup("");
+		// Reached when an object is truncated mid header, which
+		// leaves nothing at all after them. Callers render subject
+		// and msg as text without checking, so they get empty
+		// strings rather than NULL.
+		info->subject = xstrdup("");
+		info->msg = xstrdup("");
 	}
 
-	reencode(&ret->author, ret->msg_encoding, PAGE_ENCODING);
-	reencode(&ret->author_email, ret->msg_encoding, PAGE_ENCODING);
-	reencode(&ret->committer, ret->msg_encoding, PAGE_ENCODING);
-	reencode(&ret->committer_email, ret->msg_encoding, PAGE_ENCODING);
-	reencode(&ret->subject, ret->msg_encoding, PAGE_ENCODING);
-	reencode(&ret->msg, ret->msg_encoding, PAGE_ENCODING);
+	reencode(&info->author, info->msg_encoding, PAGE_ENCODING);
+	reencode(&info->author_email, info->msg_encoding, PAGE_ENCODING);
+	reencode(&info->committer, info->msg_encoding, PAGE_ENCODING);
+	reencode(&info->committer_email, info->msg_encoding, PAGE_ENCODING);
+	reencode(&info->subject, info->msg_encoding, PAGE_ENCODING);
+	reencode(&info->msg, info->msg_encoding, PAGE_ENCODING);
 
-	return ret;
+	return info;
 }
 
 struct taginfo *cgit_parse_tag(struct tag *tag)
@@ -212,18 +221,19 @@ struct taginfo *cgit_parse_tag(struct tag *tag)
 	enum object_type type;
 	unsigned long size;
 	const char *p;
-	struct taginfo *ret = NULL;
+	struct taginfo *info = NULL;
 
-	data = odb_read_object(the_repository->objects, &tag->object.oid, &type, &size);
+	data = odb_read_object(the_repository->objects, &tag->object.oid,
+			       &type, &size);
 	if (!data || type != OBJ_TAG)
 		goto cleanup;
 
-	ret = xcalloc(1, sizeof(struct taginfo));
+	info = xcalloc(1, sizeof(struct taginfo));
 
 	for (p = data; !end_of_header(p); p = next_header_line(p)) {
 		if (skip_prefix(p, "tagger ", &p)) {
-			parse_user(p, &ret->tagger, &ret->tagger_email,
-				&ret->tagger_date, &ret->tagger_tz);
+			parse_user(p, &info->tagger, &info->tagger_email,
+				&info->tagger_date, &info->tagger_tz);
 		}
 	}
 
@@ -231,9 +241,9 @@ struct taginfo *cgit_parse_tag(struct tag *tag)
 		p++;
 
 	if (p && *p)
-		ret->msg = xstrdup(p);
+		info->msg = xstrdup(p);
 
 cleanup:
 	free(data);
-	return ret;
+	return info;
 }