diff options
author | Rob Landley <rob@landley.net> | 2019-06-25 14:32:36 -0500 |
---|---|---|
committer | Rob Landley <rob@landley.net> | 2019-06-25 14:32:36 -0500 |
commit | 7fceed5f75c9c082b2bc5f47be816e69524f63af (patch) | |
tree | a8b6b18be4f944e5c2670c39474a31bccd6af0c9 | |
parent | aec194e573e6f94a7bf3c3d64b2a86c1f2057638 (diff) | |
download | toybox-7fceed5f75c9c082b2bc5f47be816e69524f63af.tar.gz |
Start over on toysh.
Basic line continuation logic (to prompt with $PS2). Doesn't use result yet.
-rw-r--r-- | toys/pending/sh.c | 484 |
1 files changed, 309 insertions, 175 deletions
diff --git a/toys/pending/sh.c b/toys/pending/sh.c index 8a9e93b6..8c898cd5 100644 --- a/toys/pending/sh.c +++ b/toys/pending/sh.c @@ -18,24 +18,6 @@ * umask unalias wait * * Things like the bash man page are good to read too. - * - * TODO: "make sh" doesn't work (nofork builtins need to be included) - * TODO: test that $PS1 color changes work without stupid \[ \] hack - * TODO: make fake pty wrapper for test infrastructure - * TODO: // Handle embedded NUL bytes in the command line. - * TODO: var=val command - * existing but considered builtins: false kill pwd true - * buitins: alias bg command fc fg getopts jobs newgrp read umask unalias wait - * "special" builtins: break continue : . eval exec export readonly return set - * shift times trap unset - * | & ; < > ( ) $ ` \ " ' <space> <tab> <newline> - * * ? [ # ~ = % - * ! { } case do done elif else esac fi for if in then until while - * [[ ]] function select - * $@ $* $# $? $- $$ $! $0 - * ENV HOME IFS LANG LC_ALL LINENO PATH PPID PS1 PS2 PS4 PWD - * label: - * TODO: test exit from "trap EXIT" doesn't recurse USE_SH(NEWTOY(cd, NULL, TOYFLAG_NOFORK)) USE_SH(NEWTOY(exit, NULL, TOYFLAG_NOFORK)) @@ -58,6 +40,7 @@ config SH -c command line to execute -i interactive mode (default when STDIN is a tty) +# These are here for the help text, they're not selectable and control nothing config CD bool default n @@ -88,25 +71,31 @@ GLOBALS( char *command; long lineno; -) -// What we know about a single process. -struct command { - struct command *next; - int flags; // exit, suspend, && || - int pid; // pid (or exit code) - int argc; - char *argv[0]; -}; - -// A collection of processes piped into/waiting on each other. -struct pipeline { - struct pipeline *next; - int job_id; - struct command *cmd; - char *cmdline; // Unparsed line for display purposes - int cmdlinelen; // How long is cmdline? -}; + // parse scratch space + struct double_list *parse; + + // Running jobs. + struct sh_job { + struct sh_job *next, *prev; + unsigned jobno; + + // Every pipeline has at least one set of arguments or it's Not A Thing + struct sh_arg { + char **v; + unsigned long c; + } pipeline; + + // null terminated array of running processes in pipeline + struct sh_process { + struct string_list *delete; // expanded strings + int pid, exit; // status? Stopped? Exited? + char *end; + struct sh_arg arg; + } *procs, *proc; + } *jobs, *job; + unsigned jobcnt; +) void cd_main(void) { @@ -120,87 +109,56 @@ void exit_main(void) exit(*toys.optargs ? atoi(*toys.optargs) : 0); } -// Parse one word from the command line, appending one or more argv[] entries -// to struct command. Handles environment variable substitution and -// substrings. Returns pointer to next used byte, or NULL if it -// hit an ending token. -static char *parse_word(char *start, struct command **cmd) -{ - char *end; - - // Detect end of line (and truncate line at comment) - if (strchr("><&|(;", *start)) return 0; - - // Grab next word. (Add dequote and envvar logic here) - end = start; - while (*end && !isspace(*end)) end++; - (*cmd)->argv[(*cmd)->argc++] = xstrndup(start, end-start); - - // Allocate more space if there's no room for NULL terminator. - - if (!((*cmd)->argc & 7)) - *cmd=xrealloc(*cmd, - sizeof(struct command) + ((*cmd)->argc+8)*sizeof(char *)); - (*cmd)->argv[(*cmd)->argc] = 0; - return end; -} - -// Parse a line of text into a pipeline. -// Returns a pointer to the next line. - -static char *parse_pipeline(char *cmdline, struct pipeline *line) +// Print prompt, parsing escapes +static void do_prompt(char *prompt) { - struct command **cmd = &(line->cmd); - char *start = line->cmdline = cmdline; - - if (!cmdline) return 0; + char *s, c, cc; - line->cmdline = cmdline; - - // Parse command into argv[] - for (;;) { - char *end; - - // Skip leading whitespace and detect end of line. - while (isspace(*start)) start++; - if (!*start || *start=='#') { - line->cmdlinelen = start-cmdline; - return 0; - } + if (!prompt) prompt = "\\$ "; + while (*prompt) { + c = *(prompt++); - // Allocate next command structure if necessary - if (!*cmd) *cmd = xzalloc(sizeof(struct command)+8*sizeof(char *)); + if (c=='!') { + if (*prompt=='!') prompt++; + else { + printf("%ld", TT.lineno); + continue; + } + } else if (c=='\\') { + int i = 0; - // Parse next argument and add the results to argv[] - end = parse_word(start, cmd); + cc = *(prompt++); + if (!cc) goto down; - // If we hit the end of this command, how did it end? - if (!end) { - if (*start) { - if (*start==';') { - start++; - break; + // \nnn \dD{}hHjlstT@AuvVwW!#$ + // Ignore bash's "nonprintable" hack; query our cursor position instead. + if (cc=='[' || cc==']') continue; + else if (cc=='$') putchar(getuid() ? '$' : '#'); + else if (cc=='h' || cc=='H') { + *toybuf = 0; + gethostname(toybuf, sizeof(toybuf)-1); + if (cc=='h' && (s = strchr(toybuf, '.'))) *s = 0; + fputs(toybuf, stdout); + } else if (cc=='s') fputs(getbasename(*toys.argv), stdout); + else { + if (!(c = unescape(cc))) { + c = '\\'; + prompt--; } - // handle | & < > >> << || && + i++; } - break; + if (!i) continue; } - start = end; +down: + putchar(c); } - - line->cmdlinelen = start-cmdline; - - return start; + fflush(stdout); } // Execute the commands in a pipeline -static void run_pipeline(struct pipeline *line) +static void run_command(struct sh_process *pp) { - struct toy_list *tl; - struct command *cmd = line->cmd; - if (!cmd || !cmd->argc) return; - - tl = toy_find(cmd->argv[0]); + struct toy_list *tl = toy_find(*pp->arg.v); // Is this command a builtin that should run in this process? if (tl && (tl->flags & TOYFLAG_NOFORK)) { @@ -213,123 +171,299 @@ static void run_pipeline(struct pipeline *line) if (!sigsetjmp(rebound, 1)) { toys.rebound = &rebound; - toy_init(tl, cmd->argv); + toy_init(tl, pp->arg.v); tl->toy_main(); } - cmd->pid = toys.exitval; + pp->exit = toys.exitval; if (toys.optargs != toys.argv+1) free(toys.optargs); if (toys.old_umask) umask(toys.old_umask); memcpy(&toys, &temp, sizeof(struct toy_context)); } else { - int status; + int pipe[2]; - cmd->pid = vfork(); - if (!cmd->pid) xexec(cmd->argv); - else waitpid(cmd->pid, &status, 0); - - if (WIFEXITED(status)) cmd->pid = WEXITSTATUS(status); - if (WIFSIGNALED(status)) cmd->pid = WTERMSIG(status); + pipe[0] = 0; + pipe[1] = 1; + if (-1 == (pp->pid = xpopen_both(pp->arg.v, pipe))) + perror_msg("%s: not found", *pp->arg.v); + else pp->exit = xpclose_both(pp->pid, 0); } return; } -// Free the contents of a command structure -static void free_cmd(void *data) +// todo: ${name:?error} causes an error/abort here (syntax_err longjmp?) +static void expand_arg(struct sh_arg *arg, char *new) { - struct command *cmd=(struct command *)data; + if (!(arg->c&32)) arg->v = xrealloc(arg->v, sizeof(void *)*(arg->c+33)); - while(cmd->argc) free(cmd->argv[--cmd->argc]); + arg->v[arg->c++] = new; + arg->v[arg->c] = 0; } - -// Parse a command line and do what it says to do. -static void handle(char *command) +// like error_msg() but exit from shell scripts +void syntax_err(char *msg, ...) { - struct pipeline line; - char *start = command; + va_list va; - // Loop through commands in this line + va_start(va, msg); + verror_msg(msg, 0, va); + va_end(va); - for (;;) { + if (*toys.optargs) xexit(); +} + + +// Parse one word from the command line, appending one or more argv[] entries +// to struct command. Handles environment variable substitution and +// substrings. Returns pointer to next used byte, or NULL if it +// hit an ending token. + +// caller eats leading spaces - // Parse a group of connected commands +// parse next word from command line. Returns end, or 0 if need continuation +static char *parse_word(char *start) +{ + int i, quote = 0; + char *end = start, *s; + + // find end of string + + while (*end) { + i = 0; + + // Handle quote contexts + if (quote) { + // end quote, skip quoted chars + if (*end == toybuf[quote-1]) quote--, end++; + else if (toybuf[quote-1] == '"' && *end == '`') toybuf[quote++] = *end++; + else if (toybuf[quote-1] == '\'' || isspace(*end)) end++; + else i++; + } else { + if (isspace(*end)) break; + // start quote + if (strchr("\"'`", *end)) toybuf[quote++] = *end++; + else if (strstart(&end, "<(") || strstart(&end,">(")) toybuf[quote++]=')'; + else if (*end==')') return end+(end==start); + else { + // control chars + for (s = end; strchr(";|&<>(", *s); s++); + if (s != end) return (end == start) ? s : end; + i++; + } + } - memset(&line,0,sizeof(struct pipeline)); - start = parse_pipeline(start, &line); - if (!line.cmd) break; + // loop if we already handled a symbol + if (!i) continue; - // Run those commands + // Things the same unquoted or in double quotes - run_pipeline(&line); - llist_traverse(line.cmd, free_cmd); + // backslash escapes + if (*end == '\\') { + if (!end[1]) return 0; + end += 2; + } else if (*end == '$') { + // barf if we're near overloading quote stack (nesting ridiculously deep) + if (quote>4000) { + syntax_err("tilt"); + return (void *)1; + } + end++; + if (strstart(&end, "((")) { + // all we care about here are parentheses matching and then ending )) + for (i = 0;;) { + if (!*end) return 0; + if (!i && strstart(&end, "))")) break; + if (*end == '(') i++; + else if (*end == ')') i--; + } + } else if (-1 != (i = stridx("({[", *end))) { + toybuf[quote++] = ")}]"[i]; + end++; + } + } else end++; } + + return quote ? 0 : end; } -static void do_prompt(void) +// Consume a line of shell script and do what it says. Returns 0 if finished, +// pointer to start of unused part of line if it needs another line of input. +static char *parse_line(char *line, struct double_list **pipeline) { - char *prompt = getenv("PS1"), *s, c, cc; + char *start = line, *end, *s; + struct sh_arg *arg = 0; + struct double_list *pl, *expect = 0; + unsigned i, paren = 0; + + // Resume appending to last pipeline's last argument list + if (*pipeline) arg = (void *)(*pipeline)->prev->data; + if (arg) for (i = 0; i<arg->c; i++) { + if (!strcmp(arg->v[i], "(")) paren++; + else if (!strcmp(arg->v[i], ")")) paren--; + } - if (!prompt) prompt = "\\$ "; - while (*prompt) { - c = *(prompt++); + // Loop handling each word + for (;;) { + // Skip leading whitespace/comment + while (isspace(*start)) ++start; + if (*start=='#') { + while (*start && *start != '\n') start++; + continue; + } - if (c=='!') { - if (*prompt=='!') prompt++; - else { - printf("%ld", TT.lineno); - continue; + // Parse next word and detect continuation/overflow. + if ((end = parse_word(start)) == (void *)1) return 0; + if (!end) return start; + + // Extend pipeline and argv[], handle EOL + if (!arg) + dlist_add(pipeline, (void *)(arg = xzalloc(sizeof(struct sh_arg)))); + if (!(31&arg->c)) arg->v = xrealloc(arg->v, (32+arg->c)*sizeof(void *)); + if (end == start) { + arg->v[arg->c] = 0; + break; + } + + // Save argument (strdup) and check if it's special + s = arg->v[arg->c] = xstrndup(start, end-start); + if (!strcmp(s, "(")) paren++; + else if (!strcmp(s, ")") && !paren--) syntax_err("bad %s", s); + if (paren || !strchr(";|&", *start)) arg->c++; + else { + if (!arg->c) { + syntax_err("bad %s", arg->v[arg->c]); + goto flush; } - } else if (c=='\\') { - cc = *(prompt++); - if (!cc) goto down; + arg = 0; + } + start = end; + } - // \nnn \dD{}hHjlstT@AuvVwW!#$ - // Ignore bash's "nonprintable" hack; query our cursor position instead. - if (cc=='[' || cc==']') continue; - else if (cc=='$') putchar(getuid() ? '$' : '#'); - else if (cc=='h' || cc=='H') { - *toybuf = 0; - gethostname(toybuf, sizeof(toybuf)-1); - if (cc=='h' && (s = strchr(toybuf, '.'))) *s = 0; - fputs(toybuf, stdout); - } else if (cc=='s') fputs(getbasename(*toys.argv), stdout); - else { - if (!(c = unescape(cc))) { - c = '\\'; - prompt--; + // We parsed to the end of the line, which ended a pipeline. + // Now handle flow control commands, which can also need more lines. + + // array of command lines separated by | and such + // Note: don't preparse past ; because environment variables differ + + // Check for flow control continuations + end = 0; + for (pl = *pipeline; pl ; pl = (pl->next == *pipeline) ? 0 : pl->next) { + arg = (void *)pl->data; + if (!arg->c) continue; + + // parse flow control statements in this command line + for (i = 0; i<arg->c; i++) { + char *ex = expect ? expect->prev->data : 0; + + s = arg->v[i]; + if (!strcmp(s, "if")) ex = "then"; + else if (!strcmp(s, "for") || !strcmp(s, "select") + || !strcmp(s, "while") || !strcmp(s, "until")) ex = "do"; + else if (!strcmp(s, "case")) ex = "esac"; + else if (!strcmp(s, "{")) ex = "}"; + else if (!strcmp(s, "[[")) ex = "]]"; + + // If we expect a non-flow-control command, eat rest of line + else if (expect && !ex) { + free(dlist_pop(&expect)); + continue; + + // Did we find a specific word we were waiting for? + } else if (ex && !strcmp(arg->v[i], ex)) { + free(dlist_pop(&expect)); + if (end && !strcmp(end, ";")) { + // can't if | then or while && do, only ; or newline counts + syntax_err("bad %s", end); + goto flush; + } + if (!strcmp(s, "do")) dlist_add(&expect, "done"); + else if (!strcmp(s, "then")) dlist_add(&expect, "fi\0A"); + break; + // fi could have elif, which queues a then. + } else if (ex && !strcmp(ex, "fi")) { + if (!strcmp(s, "elif")) { + free(dlist_pop(&expect)); + dlist_add(&expect, "then"); + // catch duplicate else while we're here + } else if (!strcmp(s, "else")) { + if (ex[3] != 'A') { + syntax_err("2 else"); + goto flush; + } + free(dlist_pop(&expect)); + dlist_add(&expect, "fi\0B"); } + } else break; - goto down; - } - continue; + dlist_add(&expect, ex); } -down: - putchar(c); + // Record how the previous stanza ended + end = arg->v[arg->c]; + } + + // If we need more lines to finish flow control... + // TODO: functions + if (expect) { + llist_traverse(expect, free); + return start; } + + for (pl = *pipeline; pl ; pl = (pl->next == *pipeline) ? 0 : pl->next) { + struct sh_process *pp = xzalloc(sizeof(struct sh_process)); + + for (i = 0; i<((struct sh_arg *)pl->data)->c; i++) + expand_arg(&pp->arg, ((struct sh_arg *)pl->data)->v[i]); + run_command(pp); + } + +flush: + while ((pl = dlist_pop(pipeline))) { + arg = (void *)pl->data; + free(pl); + for (i = 0; i<arg->c; i++) free(arg->v[i]); + free(arg->v); + free(arg); + } + *pipeline = 0; + + return 0; } void sh_main(void) { FILE *f = 0; + char *command = 0, *old = 0; + struct double_list *scratch = 0; // Set up signal handlers and grab control of this tty. if (isatty(0)) toys.optflags |= FLAG_i; if (*toys.optargs) f = xfopen(*toys.optargs, "r"); - if (TT.command) handle(xstrdup(TT.command)); - else { - size_t cmdlen = 0; - for (;;) { - char *command = 0; - - // TODO: parse escapes in prompt - if (!f) do_prompt(); - if (1 > getline(&command, &cmdlen, f ? f : stdin)) break; - handle(command); - free(command); + if (TT.command) command = parse_line(TT.command, &scratch); + else for (;;) { + char *new = 0; + size_t linelen = 0; + + // Prompt and read line + if (!f) do_prompt(getenv(command ? "PS2" : "PS1")); + if (1 > getline(&new, &linelen, f ? f : stdin)) break; + if (f) TT.lineno++; + + // Append to unused portion of previous line if any + if (command) { + command = xmprintf("%s%s", command, new); + free(old); + free(new); + old = command; + } else { + free(old); + old = new; } + + // returns 0 if line consumed, command if it needs more data + command = parse_line(old, &scratch); } - toys.exitval = 1; + if (command) error_exit("unfinished line"); + toys.exitval = f && ferror(f); } |