=================================================================== RCS file: /cvs/mandoc/cgi.c,v retrieving revision 1.59 retrieving revision 1.82 diff -u -p -r1.59 -r1.82 --- mandoc/cgi.c 2014/07/09 16:23:01 1.59 +++ mandoc/cgi.c 2014/07/24 20:30:45 1.82 @@ -1,4 +1,4 @@ -/* $Id: cgi.c,v 1.59 2014/07/09 16:23:01 schwarze Exp $ */ +/* $Id: cgi.c,v 1.82 2014/07/24 20:30:45 schwarze Exp $ */ /* * Copyright (c) 2011, 2012 Kristaps Dzonsons * Copyright (c) 2014 Ingo Schwarze @@ -33,14 +33,8 @@ #include "main.h" #include "manpath.h" #include "mansearch.h" +#include "cgi.h" -enum page { - PAGE_INDEX, - PAGE_SEARCH, - PAGE_SHOW, - PAGE__MAX -}; - /* * A query as passed to the search function. */ @@ -49,18 +43,16 @@ struct query { const char *arch; /* architecture */ const char *sec; /* manual section */ const char *expr; /* unparsed expression string */ - int legacy; /* whether legacy mode */ + int equal; /* match whole names, not substrings */ }; struct req { struct query q; char **p; /* array of available manpaths */ size_t psz; /* number of available manpaths */ - enum page page; }; static void catman(const struct req *, const char *); -static int cmp(const void *, const void *); static void format(const struct req *, const char *); static void html_print(const char *); static void html_printquery(const struct req *); @@ -71,33 +63,56 @@ static void http_print(const char *); static void http_putchar(char); static void http_printquery(const struct req *); static void pathgen(struct req *); -static void pg_index(const struct req *, char *); -static void pg_search(const struct req *, char *); -static void pg_show(const struct req *, char *); +static void pg_error_badrequest(const char *); +static void pg_error_internal(void); +static void pg_index(const struct req *); +static void pg_noresult(const struct req *, const char *); +static void pg_search(const struct req *); +static void pg_searchres(const struct req *, + struct manpage *, size_t); +static void pg_show(struct req *, const char *); static void resp_begin_html(int, const char *); static void resp_begin_http(int, const char *); static void resp_end_html(void); -static void resp_error_badrequest(const char *); -static void resp_error_internal(void); -static void resp_error_notfound(const char *); -static void resp_index(const struct req *); -static void resp_noresult(const struct req *, - const char *); -static void resp_search(const struct req *, - struct manpage *, size_t); static void resp_searchform(const struct req *); +static void resp_show(const struct req *, const char *); static const char *scriptname; /* CGI script name */ -static const char *mandir; /* contains all manpath directories */ -static const char *cssdir; /* css directory */ -static const char *httphost; /* hostname used in the URIs */ -static const char * const pages[PAGE__MAX] = { - "index", /* PAGE_INDEX */ - "search", /* PAGE_SEARCH */ - "show", /* PAGE_SHOW */ +static const int sec_prios[] = {1, 4, 5, 8, 6, 3, 7, 2, 9}; +static const char *const sec_numbers[] = { + "0", "1", "2", "3", "3p", "4", "5", "6", "7", "8", "9" }; +static const char *const sec_names[] = { + "All Sections", + "1 - General Commands", + "2 - System Calls", + "3 - Subroutines", + "3p - Perl Subroutines", + "4 - Special Files", + "5 - File Formats", + "6 - Games", + "7 - Macros and Conventions", + "8 - Maintenance Commands", + "9 - Kernel Interface" +}; +static const int sec_MAX = sizeof(sec_names) / sizeof(char *); +static const char *const arch_names[] = { + "amd64", "alpha", "armish", "armv7", + "aviion", "hppa", "hppa64", "i386", + "ia64", "landisk", "loongson", "luna88k", + "macppc", "mips64", "octeon", "sgi", + "socppc", "solbourne", "sparc", "sparc64", + "vax", "zaurus", + "amiga", "arc", "arm32", "atari", + "beagle", "cats", "hp300", "mac68k", + "mvme68k", "mvme88k", "mvmeppc", "palm", + "pc532", "pegasos", "pmax", "powerpc", + "sun3", "wgrisc", "x68k" +}; +static const int arch_MAX = sizeof(arch_names) / sizeof(char *); + /* * Print a character, escaping HTML along the way. * This will pass non-ASCII straight to output: be warned! @@ -142,9 +157,11 @@ http_printquery(const struct req *req) http_print(req->q.arch); } if (NULL != req->q.expr) { - printf("&expr="); - http_print(req->q.expr ? req->q.expr : ""); + printf("&query="); + http_print(req->q.expr); } + if (0 == req->q.equal) + printf("&apropos=1"); } static void @@ -164,9 +181,11 @@ html_printquery(const struct req *req) html_print(req->q.arch); } if (NULL != req->q.expr) { - printf("&expr="); - html_print(req->q.expr ? req->q.expr : ""); + printf("&query="); + html_print(req->q.expr); } + if (0 == req->q.equal) + printf("&apropos=1"); } static void @@ -202,12 +221,11 @@ static void http_parse(struct req *req, char *p) { char *key, *val; - int legacy; memset(&req->q, 0, sizeof(struct query)); req->q.manpath = req->p[0]; + req->q.equal = 1; - legacy = -1; while ('\0' != *p) { key = p; val = NULL; @@ -228,37 +246,32 @@ http_parse(struct req *req, char *p) if (NULL != val && ! http_decode(val)) break; - if (0 == strcmp(key, "expr")) + if (0 == strcmp(key, "query")) req->q.expr = val; - else if (0 == strcmp(key, "query")) - req->q.expr = val; - else if (0 == strcmp(key, "sec")) - req->q.sec = val; - else if (0 == strcmp(key, "sektion")) - req->q.sec = val; - else if (0 == strcmp(key, "arch")) - req->q.arch = val; - else if (0 == strcmp(key, "manpath")) + else if (0 == strcmp(key, "manpath")) { +#ifdef COMPAT_OLDURI + if (0 == strncmp(val, "OpenBSD ", 8)) { + val[7] = '-'; + if ('C' == val[8]) + val[8] = 'c'; + } +#endif req->q.manpath = val; - else if (0 == strcmp(key, "apropos")) - legacy = 0 == strcmp(val, "0"); + } else if (0 == strcmp(key, "apropos")) + req->q.equal = !strcmp(val, "0"); + else if (0 == strcmp(key, "sec")) { + if (strcmp(val, "0")) + req->q.sec = val; +#ifdef COMPAT_OLDURI + } else if (0 == strcmp(key, "sektion")) { + if (strcmp(val, "0")) + req->q.sec = val; +#endif + } else if (0 == strcmp(key, "arch")) { + if (strcmp(val, "default")) + req->q.arch = val; + } } - - /* Test for old man.cgi compatibility mode. */ - - req->q.legacy = legacy > 0; - - /* - * Section "0" means no section when in legacy mode. - * For some man.cgi scripts, "default" arch is none. - */ - - if (req->q.legacy && NULL != req->q.sec) - if (0 == strcmp(req->q.sec, "0")) - req->q.sec = NULL; - if (req->q.legacy && NULL != req->q.arch) - if (0 == strcmp(req->q.arch, "default")) - req->q.arch = NULL; } static void @@ -284,11 +297,13 @@ static int http_decode(char *p) { char hex[3]; + char *q; int c; hex[2] = '\0'; - for ( ; '\0' != *p; p++) { + q = p; + for ( ; '\0' != *p; p++, q++) { if ('%' == *p) { if ('\0' == (hex[0] = *(p + 1))) return(0); @@ -299,13 +314,13 @@ http_decode(char *p) if ('\0' == c) return(0); - *p = (char)c; - memmove(p + 1, p + 3, strlen(p + 3) + 1); + *q = (char)c; + p += 2; } else - *p = '+' == *p ? ' ' : *p; + *q = '+' == *p ? ' ' : *p; } - *p = '\0'; + *q = '\0'; return(1); } @@ -314,12 +329,12 @@ resp_begin_http(int code, const char *msg) { if (200 != code) - printf("Status: %d %s\n", code, msg); + printf("Status: %d %s\r\n", code, msg); - puts("Content-Type: text/html; charset=utf-8\n" - "Cache-Control: no-cache\n" - "Pragma: no-cache\n" - ""); + printf("Content-Type: text/html; charset=utf-8\r\n" + "Cache-Control: no-cache\r\n" + "Pragma: no-cache\r\n" + "\r\n"); fflush(stdout); } @@ -341,11 +356,11 @@ resp_begin_html(int code, const char *msg) " TYPE=\"text/css\" media=\"all\">\n" "\n" - "System Manpage Reference\n" + "%s\n" "\n" "\n" "\n", - cssdir, cssdir); + CSS_DIR, CSS_DIR, CUSTOMIZE_TITLE); } static void @@ -361,32 +376,74 @@ resp_searchform(const struct req *req) { int i; + puts(CUSTOMIZE_BEGIN); puts(""); printf("
\n" - "
\n" + "\n" "
\n" - "Search Parameters\n" - " for manuals satisfying \n" - "Manual Page Search Parameters\n", scriptname); - html_print(req->q.expr ? req->q.expr : ""); - printf("\">, section " - "q.sec ? req->q.sec : ""); - printf("\">, arch " - "q.arch ? req->q.arch : ""); - printf("\">"); + + /* Write query input box. */ + + printf( "
\n" + "q.expr) + html_print(req->q.expr); + puts("\" SIZE=\"40\">"); + + /* Write submission and reset buttons. */ + + printf( "\n" + "\n"); + + /* Write show radio button */ + + printf( "\n" + "q.equal) + printf("CHECKED "); + printf( "NAME=\"apropos\" ID=\"show\" VALUE=\"0\">\n" + "\n"); + + /* Write section selector. */ + + printf( "
\n" + ""); + + /* Write architecture selector. */ + + printf( ""); + + /* Write manpath selector. */ + if (req->psz > 1) { - puts(", "); for (i = 0; i < (int)req->psz; i++) { printf("\n" + "q.equal) + printf("CHECKED "); + printf( "NAME=\"apropos\" ID=\"search\" VALUE=\"1\">\n" + "\n"); + + puts("
\n" "
\n" "
\n" "
"); puts(""); } +static int +validate_urifrag(const char *frag) +{ + + while ('\0' != *frag) { + if ( ! (isalnum((unsigned char)*frag) || + '-' == *frag || '.' == *frag || + '/' == *frag || '_' == *frag)) + return(0); + frag++; + } + return(1); +} + +static int +validate_manpath(const struct req *req, const char* manpath) +{ + size_t i; + + if ( ! strcmp(manpath, "mandoc")) + return(1); + + for (i = 0; i < req->psz; i++) + if ( ! strcmp(manpath, req->p[i])) + return(1); + + return(0); +} + +static int +validate_filename(const char *file) +{ + + if ('.' == file[0] && '/' == file[1]) + file += 2; + + return ( ! (strstr(file, "../") || strstr(file, "/..") || + (strncmp(file, "man", 3) && strncmp(file, "cat", 3)))); +} + static void -resp_index(const struct req *req) +pg_index(const struct req *req) { resp_begin_html(200, NULL); resp_searchform(req); + printf("

\n" + "This web interface is documented in the " + "man.cgi " + "manual, and the " + "apropos " + "manual explains the query syntax.\n" + "

\n", + scriptname, scriptname); resp_end_html(); } static void -resp_noresult(const struct req *req, const char *msg) +pg_noresult(const struct req *req, const char *msg) { resp_begin_html(200, NULL); resp_searchform(req); @@ -424,7 +538,7 @@ resp_noresult(const struct req *req, const char *msg) } static void -resp_error_badrequest(const char *msg) +pg_error_badrequest(const char *msg) { resp_begin_html(400, "Bad Request"); @@ -438,52 +552,46 @@ resp_error_badrequest(const char *msg) } static void -resp_error_notfound(const char *page) +pg_error_internal(void) { - - resp_begin_html(404, "Not Found"); - puts("

Page Not Found

\n" - "

\n" - "The page you're looking for, "); - printf(""); - html_print(page); - printf(",\n" - "could not be found.\n" - "Try searching from the\n" - "main page.\n" - "

", scriptname); - resp_end_html(); -} - -static void -resp_error_internal(void) -{ resp_begin_html(500, "Internal Server Error"); puts("

Internal Server Error

"); resp_end_html(); } static void -resp_search(const struct req *req, struct manpage *r, size_t sz) +pg_searchres(const struct req *req, struct manpage *r, size_t sz) { - size_t i; + char *arch, *archend; + size_t i, iuse, isec; + int archprio, archpriouse; + int prio, priouse; + char sec; + for (i = 0; i < sz; i++) { + if (validate_filename(r[i].file)) + continue; + fprintf(stderr, "invalid filename %s in %s database\n", + r[i].file, req->q.manpath); + pg_error_internal(); + return; + } + if (1 == sz) { /* * If we have just one result, then jump there now * without any delay. */ - puts("Status: 303 See Other"); - printf("Location: http://%s%s/show/%s/%s?", - httphost, scriptname, req->q.manpath, r[0].file); + printf("Status: 303 See Other\r\n"); + printf("Location: http://%s%s/%s/%s?", + HTTP_HOST, scriptname, req->q.manpath, r[0].file); http_printquery(req); - puts("\n" - "Content-Type: text/html; charset=utf-8\n"); + printf("\r\n" + "Content-Type: text/html; charset=utf-8\r\n" + "\r\n"); return; } - qsort(r, sz, sizeof(struct manpage), cmp); - resp_begin_html(200, NULL); resp_searchform(req); puts("
"); @@ -492,7 +600,7 @@ resp_search(const struct req *req, struct manpage *r, for (i = 0; i < sz; i++) { printf("\n" "\n" - "q.manpath, r[i].file); html_printquery(req); printf("\">"); @@ -507,15 +615,49 @@ resp_search(const struct req *req, struct manpage *r, puts("\n" "
"); - resp_end_html(); -} -/* ARGSUSED */ -static void -pg_index(const struct req *req, char *path) -{ + /* + * In man(1) mode, show one of the pages + * even if more than one is found. + */ - resp_index(req); + if (req->q.equal) { + puts("
"); + iuse = 0; + priouse = 10; + archpriouse = 3; + for (i = 0; i < sz; i++) { + isec = strcspn(r[i].file, "123456789"); + sec = r[i].file[isec]; + if ('\0' == sec) + continue; + prio = sec_prios[sec - '1']; + if (NULL == req->q.arch) { + archprio = + (NULL == (arch = strchr( + r[i].file + isec, '/'))) ? 3 : + (NULL == (archend = strchr( + arch + 1, '/'))) ? 0 : + strncmp(arch, "amd64/", + archend - arch) ? 2 : 1; + if (archprio < archpriouse) { + archpriouse = archprio; + priouse = prio; + iuse = i; + continue; + } + if (archprio > archpriouse) + continue; + } + if (prio >= priouse) + continue; + priouse = prio; + iuse = i; + } + resp_show(req, r[iuse].file); + } + + resp_end_html(); } static void @@ -528,13 +670,10 @@ catman(const struct req *req, const char *file) int italic, bold; if (NULL == (f = fopen(file, "r"))) { - resp_error_badrequest( - "You specified an invalid manual file."); + puts("

You specified an invalid manual file.

"); return; } - resp_begin_html(200, NULL); - resp_searchform(req); puts("
\n" "
");
 
@@ -648,9 +787,7 @@ catman(const struct req *req, const char *file)
 	}
 
 	puts("
\n" - "
\n" - "\n" - ""); + ""); fclose(f); } @@ -667,8 +804,7 @@ format(const struct req *req, const char *file) char opts[PATH_MAX + 128]; if (-1 == (fd = open(file, O_RDONLY, 0))) { - resp_error_badrequest( - "You specified an invalid manual file."); + puts("

You specified an invalid manual file.

"); return; } @@ -680,26 +816,24 @@ format(const struct req *req, const char *file) if (rc >= MANDOCLEVEL_FATAL) { fprintf(stderr, "fatal mandoc error: %s/%s\n", req->q.manpath, file); - resp_error_internal(); + pg_error_internal(); return; } - snprintf(opts, sizeof(opts), - "fragment,man=%s/search?sec=%%S&expr=Nm~^%%N$", - scriptname); + snprintf(opts, sizeof(opts), "fragment,man=%s?" + "manpath=%s&query=%%N&sec=%%S&arch=%s", + scriptname, req->q.manpath, + req->q.arch ? req->q.arch : ""); mparse_result(mp, &mdoc, &man, NULL); if (NULL == man && NULL == mdoc) { fprintf(stderr, "fatal mandoc error: %s/%s\n", req->q.manpath, file); - resp_error_internal(); + pg_error_internal(); mparse_free(mp); return; } - resp_begin_html(200, NULL); - resp_searchform(req); - vp = html_alloc(opts); if (NULL != mdoc) @@ -707,25 +841,41 @@ format(const struct req *req, const char *file) else html_man(vp, man); - puts("\n" - ""); - html_free(vp); mparse_free(mp); } static void -pg_show(const struct req *req, char *path) +resp_show(const struct req *req, const char *file) { + + if ('.' == file[0] && '/' == file[1]) + file += 2; + + if ('c' == *file) + catman(req, file); + else + format(req, file); +} + +static void +pg_show(struct req *req, const char *path) +{ char *sub; if (NULL == path || NULL == (sub = strchr(path, '/'))) { - resp_error_badrequest( + pg_error_badrequest( "You did not specify a page to show."); return; } *sub++ = '\0'; + if ( ! validate_manpath(req, path)) { + pg_error_badrequest( + "You specified an invalid manpath."); + return; + } + /* * Begin by chdir()ing into the manpath. * This way we can pick up the database files, which are @@ -733,19 +883,29 @@ pg_show(const struct req *req, char *path) */ if (-1 == chdir(path)) { - resp_error_badrequest( - "You specified an invalid manpath."); + fprintf(stderr, "chdir %s: %s\n", + path, strerror(errno)); + pg_error_internal(); return; } - if ('c' == *sub) - catman(req, sub); - else - format(req, sub); + if ( ! validate_filename(sub)) { + pg_error_badrequest( + "You specified an invalid manual file."); + return; + } + + if (strcmp(path, "mandoc")) + req->q.manpath = path; + + resp_begin_html(200, NULL); + resp_searchform(req); + resp_show(req, sub); + resp_end_html(); } static void -pg_search(const struct req *req, char *path) +pg_search(const struct req *req) { struct mansearch search; struct manpaths paths; @@ -762,15 +922,16 @@ pg_search(const struct req *req, char *path) */ if (-1 == (chdir(req->q.manpath))) { - resp_error_badrequest( - "You specified an invalid manpath."); + fprintf(stderr, "chdir %s: %s\n", + req->q.manpath, strerror(errno)); + pg_error_internal(); return; } search.arch = req->q.arch; search.sec = req->q.sec; - search.deftype = TYPE_Nm | TYPE_Nd; - search.flags = 0; + search.deftype = req->q.equal ? TYPE_Nm : (TYPE_Nm | TYPE_Nd); + search.flags = req->q.equal ? MANSEARCH_MAN : 0; paths.sz = 1; paths.paths = mandoc_malloc(sizeof(char *)); @@ -800,11 +961,11 @@ pg_search(const struct req *req, char *path) } if (0 == mansearch(&search, &paths, sz, cp, "Nd", &res, &ressz)) - resp_noresult(req, "You entered an invalid query."); + pg_noresult(req, "You entered an invalid query."); else if (0 == ressz) - resp_noresult(req, "No results found."); + pg_noresult(req, "No results found."); else - resp_search(req, res, ressz); + pg_searchres(req, res, ressz); for (i = 0; i < sz; i++) free(cp[i]); @@ -824,34 +985,33 @@ pg_search(const struct req *req, char *path) int main(void) { - int i; struct req req; - char *querystring, *path, *subpath; + const char *path; + char *querystring; + int i; /* Scan our run-time environment. */ - if (NULL == (mandir = getenv("MAN_DIR"))) - mandir = "/man"; - if (NULL == (scriptname = getenv("SCRIPT_NAME"))) scriptname = ""; - if (NULL == (cssdir = getenv("CSS_DIR"))) - cssdir = ""; + if ( ! validate_urifrag(scriptname)) { + fprintf(stderr, "unsafe SCRIPT_NAME \"%s\"\n", + scriptname); + pg_error_internal(); + return(EXIT_FAILURE); + } - if (NULL == (httphost = getenv("HTTP_HOST"))) - httphost = "localhost"; - /* - * First we change directory into the mandir so that + * First we change directory into the MAN_DIR so that * subsequent scanning for manpath directories is rooted * relative to the same position. */ - if (-1 == chdir(mandir)) { + if (-1 == chdir(MAN_DIR)) { fprintf(stderr, "MAN_DIR: %s: %s\n", - mandir, strerror(errno)); - resp_error_internal(); + MAN_DIR, strerror(errno)); + pg_error_internal(); return(EXIT_FAILURE); } @@ -863,67 +1023,39 @@ main(void) if (NULL != (querystring = getenv("QUERY_STRING"))) http_parse(&req, querystring); - /* - * Now juggle paths to extract information. - * We want to extract our filetype (the file suffix), the - * initial path component, then the trailing component(s). - * Start with leading subpath component. - */ + if ( ! validate_manpath(&req, req.q.manpath)) { + pg_error_badrequest( + "You specified an invalid manpath."); + return(EXIT_FAILURE); + } - subpath = path = NULL; - req.page = PAGE__MAX; + if ( ! (NULL == req.q.arch || validate_urifrag(req.q.arch))) { + pg_error_badrequest( + "You specified an invalid architecture."); + return(EXIT_FAILURE); + } - if (NULL == (path = getenv("PATH_INFO")) || '\0' == *path) - req.page = PAGE_INDEX; + /* Dispatch to the three different pages. */ - if (NULL != path && '/' == *path && '\0' == *++path) - req.page = PAGE_INDEX; + path = getenv("PATH_INFO"); + if (NULL == path) + path = ""; + else if ('/' == *path) + path++; - /* Resolve subpath component. */ + if ('\0' != *path) + pg_show(&req, path); + else if (NULL != req.q.expr) + pg_search(&req); + else + pg_index(&req); - if (NULL != path && NULL != (subpath = strchr(path, '/'))) - *subpath++ = '\0'; - - /* Map path into one we recognise. */ - - if (NULL != path && '\0' != *path) - for (i = 0; i < (int)PAGE__MAX; i++) - if (0 == strcmp(pages[i], path)) { - req.page = (enum page)i; - break; - } - - /* Route pages. */ - - switch (req.page) { - case (PAGE_INDEX): - pg_index(&req, subpath); - break; - case (PAGE_SEARCH): - pg_search(&req, subpath); - break; - case (PAGE_SHOW): - pg_show(&req, subpath); - break; - default: - resp_error_notfound(path); - break; - } - for (i = 0; i < (int)req.psz; i++) free(req.p[i]); free(req.p); return(EXIT_SUCCESS); } -static int -cmp(const void *p1, const void *p2) -{ - - return(strcasecmp(((const struct manpage *)p1)->names, - ((const struct manpage *)p2)->names)); -} - /* * Scan for indexable paths. */ @@ -934,14 +1066,37 @@ pathgen(struct req *req) char *dp; size_t dpsz; - if (NULL == (fp = fopen("manpath.conf", "r"))) - return; + if (NULL == (fp = fopen("manpath.conf", "r"))) { + fprintf(stderr, "%s/manpath.conf: %s\n", + MAN_DIR, strerror(errno)); + pg_error_internal(); + exit(EXIT_FAILURE); + } while (NULL != (dp = fgetln(fp, &dpsz))) { if ('\n' == dp[dpsz - 1]) dpsz--; req->p = mandoc_realloc(req->p, (req->psz + 1) * sizeof(char *)); - req->p[req->psz++] = mandoc_strndup(dp, dpsz); + dp = mandoc_strndup(dp, dpsz); + if ( ! validate_urifrag(dp)) { + fprintf(stderr, "%s/manpath.conf contains " + "unsafe path \"%s\"\n", MAN_DIR, dp); + pg_error_internal(); + exit(EXIT_FAILURE); + } + if (NULL != strchr(dp, '/')) { + fprintf(stderr, "%s/manpath.conf contains " + "path with slash \"%s\"\n", MAN_DIR, dp); + pg_error_internal(); + exit(EXIT_FAILURE); + } + req->p[req->psz++] = dp; + } + + if ( req->p == NULL ) { + fprintf(stderr, "%s/manpath.conf is empty\n", MAN_DIR); + pg_error_internal(); + exit(EXIT_FAILURE); } }