| /* ui-plain.c: functions for output of plain blobs by path |
| * |
| * Copyright (C) 2008 Lars Hjemli |
| * |
| * Licensed under GNU General Public License v2 |
| * (see COPYING for full license text) |
| */ |
| |
| #include <stdio.h> |
| #include "cgit.h" |
| #include "html.h" |
| #include "ui-shared.h" |
| |
| int match_baselen; |
| int match; |
| |
| static char *get_mimetype_from_file(const char *filename, const char *ext) |
| { |
| static const char *delimiters; |
| char *result; |
| FILE *fd; |
| char line[1024]; |
| char *mimetype; |
| char *token; |
| |
| if (!filename) |
| return NULL; |
| |
| fd = fopen(filename, "r"); |
| if (!fd) |
| return NULL; |
| |
| delimiters = " \t\r\n"; |
| result = NULL; |
| |
| /* loop over all lines in the file */ |
| while (!result && fgets(line, sizeof(line), fd)) { |
| mimetype = strtok(line, delimiters); |
| |
| /* skip empty lines and comment lines */ |
| if (!mimetype || (mimetype[0] == '#')) |
| continue; |
| |
| /* loop over all extensions of mimetype */ |
| while ((token = strtok(NULL, delimiters))) { |
| if (!strcasecmp(ext, token)) { |
| result = xstrdup(mimetype); |
| break; |
| } |
| } |
| } |
| fclose(fd); |
| |
| return result; |
| } |
| |
| static void print_object(const unsigned char *sha1, const char *path) |
| { |
| enum object_type type; |
| char *buf, *ext; |
| unsigned long size; |
| struct string_list_item *mime; |
| int freemime; |
| |
| type = sha1_object_info(sha1, &size); |
| if (type == OBJ_BAD) { |
| html_status(404, "Not found", 0); |
| return; |
| } |
| |
| buf = read_sha1_file(sha1, &type, &size); |
| if (!buf) { |
| html_status(404, "Not found", 0); |
| return; |
| } |
| ctx.page.mimetype = NULL; |
| ext = strrchr(path, '.'); |
| freemime = 0; |
| if (ext && *(++ext)) { |
| mime = string_list_lookup(&ctx.cfg.mimetypes, ext); |
| if (mime) { |
| ctx.page.mimetype = (char *)mime->util; |
| } else { |
| ctx.page.mimetype = get_mimetype_from_file(ctx.cfg.mimetype_file, ext); |
| if (ctx.page.mimetype) |
| freemime = 1; |
| } |
| } |
| if (!ctx.page.mimetype) { |
| if (buffer_is_binary(buf, size)) |
| ctx.page.mimetype = "application/octet-stream"; |
| else |
| ctx.page.mimetype = "text/plain"; |
| } |
| ctx.page.filename = fmt("%s", path); |
| ctx.page.size = size; |
| ctx.page.etag = sha1_to_hex(sha1); |
| cgit_print_http_headers(&ctx); |
| html_raw(buf, size); |
| match = 1; |
| if (freemime) |
| free(ctx.page.mimetype); |
| } |
| |
| static char *buildpath(const char *base, int baselen, const char *path) |
| { |
| if (path[0]) |
| return fmt("%.*s%s/", baselen, base, path); |
| else |
| return fmt("%.*s/", baselen, base); |
| } |
| |
| static void print_dir(const unsigned char *sha1, const char *base, |
| int baselen, const char *path) |
| { |
| char *fullpath, *slash; |
| size_t len; |
| |
| fullpath = buildpath(base, baselen, path); |
| slash = (fullpath[0] == '/' ? "" : "/"); |
| ctx.page.etag = sha1_to_hex(sha1); |
| cgit_print_http_headers(&ctx); |
| htmlf("<html><head><title>%s", slash); |
| html_txt(fullpath); |
| htmlf("</title></head>\n<body>\n<h2>%s", slash); |
| html_txt(fullpath); |
| html("</h2>\n<ul>\n"); |
| len = strlen(fullpath); |
| if (len > 1) { |
| fullpath[len - 1] = 0; |
| slash = strrchr(fullpath, '/'); |
| if (slash) |
| *(slash + 1) = 0; |
| else |
| fullpath = NULL; |
| html("<li>"); |
| cgit_plain_link("../", NULL, NULL, ctx.qry.head, ctx.qry.sha1, |
| fullpath); |
| html("</li>\n"); |
| } |
| match = 2; |
| } |
| |
| static void print_dir_entry(const unsigned char *sha1, const char *base, |
| int baselen, const char *path, unsigned mode) |
| { |
| char *fullpath; |
| |
| fullpath = buildpath(base, baselen, path); |
| if (!S_ISDIR(mode) && !S_ISGITLINK(mode)) |
| fullpath[strlen(fullpath) - 1] = 0; |
| html(" <li>"); |
| if (S_ISGITLINK(mode)) { |
| cgit_submodule_link(NULL, fullpath, sha1_to_hex(sha1)); |
| } else |
| cgit_plain_link(path, NULL, NULL, ctx.qry.head, ctx.qry.sha1, |
| fullpath); |
| html("</li>\n"); |
| match = 2; |
| } |
| |
| static void print_dir_tail(void) |
| { |
| html(" </ul>\n</body></html>\n"); |
| } |
| |
| static int walk_tree(const unsigned char *sha1, const char *base, int baselen, |
| const char *pathname, unsigned mode, int stage, |
| void *cbdata) |
| { |
| if (baselen == match_baselen) { |
| if (S_ISREG(mode)) |
| print_object(sha1, pathname); |
| else if (S_ISDIR(mode)) { |
| print_dir(sha1, base, baselen, pathname); |
| return READ_TREE_RECURSIVE; |
| } |
| } |
| else if (baselen > match_baselen) |
| print_dir_entry(sha1, base, baselen, pathname, mode); |
| else if (S_ISDIR(mode)) |
| return READ_TREE_RECURSIVE; |
| |
| return 0; |
| } |
| |
| static int basedir_len(const char *path) |
| { |
| char *p = strrchr(path, '/'); |
| if (p) |
| return p - path + 1; |
| return 0; |
| } |
| |
| void cgit_print_plain(struct cgit_context *ctx) |
| { |
| const char *rev = ctx->qry.sha1; |
| unsigned char sha1[20]; |
| struct commit *commit; |
| const char *paths[] = {ctx->qry.path, NULL}; |
| |
| if (!rev) |
| rev = ctx->qry.head; |
| |
| if (get_sha1(rev, sha1)) { |
| html_status(404, "Not found", 0); |
| return; |
| } |
| commit = lookup_commit_reference(sha1); |
| if (!commit || parse_commit(commit)) { |
| html_status(404, "Not found", 0); |
| return; |
| } |
| if (!paths[0]) { |
| paths[0] = ""; |
| match_baselen = -1; |
| print_dir(commit->tree->object.sha1, "", 0, ""); |
| } |
| else |
| match_baselen = basedir_len(paths[0]); |
| read_tree_recursive(commit->tree, "", 0, 0, paths, walk_tree, NULL); |
| if (!match) |
| html_status(404, "Not found", 0); |
| else if (match == 2) |
| print_dir_tail(); |
| } |