From 8f8c4cdd176fda4d93a2d1d4b0ae6321d5706e5f Mon Sep 17 00:00:00 2001
From: Kevin Ballard <kevin@sb.org>
Date: Sun, 21 Sep 2014 19:18:56 -0700
Subject: [PATCH] Implement new `read --null` flag

The `--null` flag to `read` makes it split incoming lines on NUL instead
of newlines. This is intended for processing the output of a command
that uses NUL separators (such as `find -print0`).

Fixes #1694.
---
 Makefile.in      | 14 ++++++---
 builtin.cpp      | 80 +++++++++++++++++++++++++-----------------------
 doc_src/read.txt |  2 ++
 fish_tests.cpp   | 33 ++++++++++++++++++++
 tests/read.in    | 25 +++++++++++++++
 tests/read.out   | 12 ++++++++
 wcstringutil.cpp | 40 ++++++++++++++++++++++++
 wcstringutil.h   | 29 ++++++++++++++++++
 8 files changed, 192 insertions(+), 43 deletions(-)
 create mode 100644 wcstringutil.cpp
 create mode 100644 wcstringutil.h

diff --git a/Makefile.in b/Makefile.in
index fea9ae5bf..d6d1d09dc 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -91,7 +91,7 @@ FISH_OBJS := function.o builtin.o complete.o env.o exec.o expand.o		\
 	signal.o io.o parse_util.o common.o screen.o path.o autoload.o		\
 	parser_keywords.o iothread.o color.o postfork.o	\
 	builtin_test.o parse_tree.o parse_productions.o parse_execution.o \
-	pager.o utf8.o fish_version.o
+	pager.o utf8.o fish_version.o wcstringutil.o
 
 FISH_INDENT_OBJS := fish_indent.o print_help.o common.o	\
 parser_keywords.o wutil.o tokenizer.o fish_version.o
@@ -843,9 +843,10 @@ builtin.o: io.h function.h event.h complete.h proc.h parse_tree.h tokenizer.h
 builtin.o: parse_constants.h parser.h reader.h highlight.h env.h color.h
 builtin.o: wgetopt.h sanity.h wildcard.h expand.h input_common.h input.h
 builtin.o: intern.h exec.h parse_util.h autoload.h lru.h parser_keywords.h
-builtin.o: path.h history.h builtin_set.cpp builtin_commandline.cpp
-builtin.o: builtin_complete.cpp builtin_ulimit.cpp builtin_jobs.cpp
-builtin.o: builtin_set_color.cpp output.h screen.h builtin_printf.cpp
+builtin.o: path.h history.h wcstringutil.h builtin_set.cpp
+builtin.o: builtin_commandline.cpp builtin_complete.cpp builtin_ulimit.cpp
+builtin.o: builtin_jobs.cpp builtin_set_color.cpp output.h screen.h
+builtin.o: builtin_printf.cpp
 builtin_commandline.o: config.h signal.h fallback.h util.h wutil.h common.h
 builtin_commandline.o: builtin.h io.h wgetopt.h reader.h complete.h
 builtin_commandline.o: highlight.h env.h color.h proc.h parse_tree.h
@@ -860,6 +861,7 @@ builtin_jobs.o: config.h fallback.h signal.h util.h wutil.h common.h
 builtin_jobs.o: builtin.h io.h proc.h parse_tree.h tokenizer.h
 builtin_jobs.o: parse_constants.h parser.h event.h function.h wgetopt.h
 builtin_printf.o: common.h util.h
+builtin_scripts.o: builtin_scripts.h
 builtin_set.o: config.h signal.h fallback.h util.h wutil.h common.h builtin.h
 builtin_set.o: io.h env.h expand.h parse_constants.h wgetopt.h proc.h
 builtin_set.o: parse_tree.h tokenizer.h parser.h event.h function.h
@@ -911,6 +913,7 @@ fish_tests.o: highlight.h env.h color.h builtin.h function.h event.h
 fish_tests.o: autoload.h lru.h wutil.h expand.h parser.h output.h screen.h
 fish_tests.o: exec.h path.h history.h iothread.h postfork.h parse_util.h
 fish_tests.o: pager.h input.h input_common.h utf8.h env_universal_common.h
+fish_tests.o: wcstringutil.h
 fish_version.o: fish_version.h
 function.o: config.h signal.h wutil.h common.h util.h fallback.h function.h
 function.o: event.h proc.h io.h parse_tree.h tokenizer.h parse_constants.h
@@ -930,7 +933,7 @@ input.o: complete.h highlight.h env.h color.h proc.h parse_tree.h tokenizer.h
 input.o: parse_constants.h sanity.h input_common.h input.h parser.h event.h
 input.o: function.h expand.h output.h screen.h intern.h
 input_common.o: config.h fallback.h signal.h util.h common.h wutil.h
-input_common.o: input_common.h iothread.h
+input_common.o: input_common.h env_universal_common.h env.h iothread.h
 intern.o: config.h fallback.h signal.h util.h wutil.h common.h intern.h
 io.o: config.h fallback.h signal.h util.h wutil.h common.h exec.h proc.h io.h
 io.o: parse_tree.h tokenizer.h parse_constants.h
@@ -993,6 +996,7 @@ signal.o: parse_tree.h tokenizer.h parse_constants.h
 tokenizer.o: config.h fallback.h signal.h util.h wutil.h common.h tokenizer.h
 utf8.o: utf8.h
 util.o: config.h fallback.h signal.h util.h common.h wutil.h
+wcstringutil.o: config.h wcstringutil.h common.h util.h
 wgetopt.o: config.h wgetopt.h wutil.h common.h util.h fallback.h signal.h
 wildcard.o: config.h fallback.h signal.h util.h wutil.h common.h complete.h
 wildcard.o: wildcard.h expand.h parse_constants.h reader.h io.h highlight.h
diff --git a/builtin.cpp b/builtin.cpp
index 3ab3bb486..4604190c3 100644
--- a/builtin.cpp
+++ b/builtin.cpp
@@ -65,6 +65,7 @@
 #include "path.h"
 #include "history.h"
 #include "parse_tree.h"
+#include "wcstringutil.h"
 
 /**
    The default prompt for the read command
@@ -2305,10 +2306,9 @@ static int builtin_random(parser_t &parser, wchar_t **argv)
 */
 static int builtin_read(parser_t &parser, wchar_t **argv)
 {
-    wchar_t *buff=0;
+    wcstring buff;
     int i, argc = builtin_count_args(argv);
     int place = ENV_USER;
-    wchar_t *nxt;
     const wchar_t *prompt = DEFAULT_READ_PROMPT;
     const wchar_t *commandline = L"";
     int exit_res=STATUS_BUILTIN_OK;
@@ -2317,6 +2317,7 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
     wchar_t *end;
     int shell = 0;
     int array = 0;
+    bool split_null = false;
 
     woptind=0;
 
@@ -2369,6 +2370,10 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
                 L"array", no_argument, 0, 'a'
             }
             ,
+            {
+                L"null", no_argument, 0, 'z'
+            }
+            ,
             {
                 L"help", no_argument, 0, 'h'
             }
@@ -2383,7 +2388,7 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
 
         int opt = wgetopt_long(argc,
                                argv,
-                               L"xglUup:c:hm:n:sa",
+                               L"xglUup:c:hm:n:saz",
                                long_options,
                                &opt_index);
         if (opt == -1)
@@ -2468,6 +2473,10 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
                 array = 1;
                 break;
 
+            case L'z':
+                split_null = true;
+                break;
+
             case 'h':
                 builtin_print_help(parser, argv[0], stdout_buffer);
                 return STATUS_BUILTIN_OK;
@@ -2541,7 +2550,7 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
     /*
       Check if we should read interactively using \c reader_readline()
     */
-    if (isatty(0) && builtin_stdin == 0)
+    if (isatty(0) && builtin_stdin == 0 && !split_null)
     {
         const wchar_t *line;
 
@@ -2572,13 +2581,11 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
                 // note: we're deliberately throwing away the tail of the commandline.
                 // It shouldn't be unread because it was produced with `commandline -i`,
                 // not typed.
-                buff = (wchar_t *)malloc(((size_t)nchars + 1) * sizeof(wchar_t));
-                wmemcpy(buff, line, (size_t)nchars);
-                buff[nchars] = 0;
+                buff = wcstring(line, nchars);
             }
             else
             {
-                buff = wcsdup(line);
+                buff = wcstring(line);
             }
         }
         else
@@ -2591,7 +2598,7 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
     {
         int eof=0;
 
-        wcstring sb;
+        buff.clear();
 
         while (1)
         {
@@ -2621,7 +2628,6 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
                     case (size_t)(-2):
                         break;
                     case 0:
-                        eof=1;
                         finished = 1;
                         break;
 
@@ -2635,44 +2641,43 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
             if (eof)
                 break;
 
-            if (res == L'\n')
+            if (!split_null && res == L'\n')
                 break;
 
-            sb.push_back(res);
+            if (split_null && res == L'\0')
+                break;
 
-            if (0 < nchars && (size_t)nchars <= sb.size())
+            buff.push_back(res);
+
+            if (0 < nchars && (size_t)nchars <= buff.size())
             {
                 break;
             }
         }
 
-        if (sb.size() < 2 && eof)
+        if (buff.size() < 2 && eof)
         {
             exit_res = 1;
         }
-
-        buff = wcsdup(sb.c_str());
     }
 
     if (i != argc && !exit_res)
     {
-
-        wchar_t *state;
-
         env_var_t ifs = env_get_string(L"IFS");
-
         if (ifs.missing_or_empty())
         {
             /* Every character is a separate token */
-            size_t bufflen = wcslen(buff);
+            size_t bufflen = buff.size();
             if (array)
             {
                 if (bufflen > 0)
                 {
                     wcstring chars(bufflen+(bufflen-1), ARRAY_SEP);
-                    for (size_t j=0; j<bufflen; ++j)
+                    wcstring::iterator out = chars.begin();
+                    for (wcstring::const_iterator it = buff.begin(), end = buff.end(); it != end; ++it)
                     {
-                        chars[j*2] = buff[j];
+                        *out = *it;
+                        out += 2;
                     }
                     env_set(argv[i], chars.c_str(), place);
                 }
@@ -2686,14 +2691,15 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
                 size_t j = 0;
                 for (; i+1 < argc; ++i)
                 {
-                    if (j < bufflen) {
-                        wchar_t buffer[2] = {buff[j], 0};
+                    if (j < bufflen)
+                    {
+                        wchar_t buffer[2] = {buff[j++], 0};
                         env_set(argv[i], buffer, place);
                     }
-                    else {
+                    else
+                    {
                         env_set(argv[i], L"", place);
                     }
-                    if (j < bufflen) ++j;
                 }
                 if (i < argc) env_set(argv[i], &buff[j], place);
             }
@@ -2701,34 +2707,32 @@ static int builtin_read(parser_t &parser, wchar_t **argv)
         else if (array)
         {
             wcstring tokens;
-            tokens.reserve(wcslen(buff));
+            tokens.reserve(buff.size());
             bool empty = true;
 
-            for (nxt = wcstok(buff, ifs.c_str(), &state); nxt != 0; nxt = wcstok(0, ifs.c_str(), &state))
+            for (wcstring_range loc = wcstring_tok(buff, ifs); loc.first != wcstring::npos; loc = wcstring_tok(buff, ifs, loc))
             {
-                if (! tokens.empty()) tokens.push_back(ARRAY_SEP);
-                tokens.append(nxt);
+                if (!empty) tokens.push_back(ARRAY_SEP);
+                tokens.append(buff, loc.first, loc.second);
                 empty = false;
             }
             env_set(argv[i], empty ? NULL : tokens.c_str(), place);
         }
         else
         {
-            nxt = wcstok(buff, (i<argc-1)?ifs.c_str():L"", &state);
+            wcstring_range loc = wcstring_range(0,0);
 
             while (i<argc)
             {
-                env_set(argv[i], nxt != 0 ? nxt: L"", place);
+                loc = wcstring_tok(buff, (i+1<argc) ? ifs : L"", loc);
+                env_set(argv[i], loc.first == wcstring::npos ? L"" : &buff.c_str()[loc.first], place);
 
-                i++;
-                if (nxt != 0)
-                    nxt = wcstok(0, (i<argc-1)?ifs.c_str():L"", &state);
+                ++i;
             }
+
         }
     }
 
-    free(buff);
-
     return exit_res;
 }
 
diff --git a/doc_src/read.txt b/doc_src/read.txt
index aee3e85b8..79335a794 100644
--- a/doc_src/read.txt
+++ b/doc_src/read.txt
@@ -33,6 +33,8 @@ The following options are available:
 
 - `-a` or `--array` stores the result as an array.
 
+- `-z` or `--null` reads up to NUL instead of newline. Disables interactive mode.
+
 `read` reads a single line of input from stdin, breaks it into tokens based on the `IFS` shell variable, and then assigns one token to each variable specified in `VARIABLES`. If there are more tokens than variables, the complete remainder is assigned to the last variable. As a special case, if `IFS` is set to the empty string, each character of the input is considered a separate token.
 
 If `-a` or `--array` is provided, only one variable name is allowed and the tokens are stored as an array in this variable.
diff --git a/fish_tests.cpp b/fish_tests.cpp
index 744da79d9..7222065f2 100644
--- a/fish_tests.cpp
+++ b/fish_tests.cpp
@@ -65,6 +65,7 @@
 #include "input.h"
 #include "utf8.h"
 #include "env_universal_common.h"
+#include "wcstringutil.h"
 
 static const char * const * s_arguments;
 static int s_test_run_count = 0;
@@ -3629,6 +3630,37 @@ static void test_highlighting(void)
     }
 }
 
+static void test_wcstring_tok(void)
+{
+    say(L"Testing wcstring_tok");
+    wcstring buff = L"hello world";
+    wcstring needle = L" \t\n";
+    wcstring_range loc = wcstring_tok(buff, needle);
+    if (loc.first == wcstring::npos || buff.substr(loc.first, loc.second) != L"hello")
+    {
+        err(L"Wrong results from first wcstring_tok(): {%zu, %zu}", loc.first, loc.second);
+    }
+    loc = wcstring_tok(buff, needle, loc);
+    if (loc.first == wcstring::npos || buff.substr(loc.first, loc.second) != L"world")
+    {
+        err(L"Wrong results from second wcstring_tok(): {%zu, %zu}", loc.first, loc.second);
+    }
+    loc = wcstring_tok(buff, needle, loc);
+    if (loc.first != wcstring::npos)
+    {
+        err(L"Wrong results from third wcstring_tok(): {%zu, %zu}", loc.first, loc.second);
+    }
+
+    buff = L"hello world";
+    loc = wcstring_tok(buff, needle);
+    // loc is "hello" again
+    loc = wcstring_tok(buff, L"", loc);
+    if (loc.first == wcstring::npos || buff.substr(loc.first, loc.second) != L"world")
+    {
+        err(L"Wrong results from wcstring_tok with empty needle: {%zu, %zu}", loc.first, loc.second);
+    }
+}
+
 /**
    Main test
 */
@@ -3709,6 +3741,7 @@ int main(int argc, char **argv)
     if (should_test_function("autosuggestion_ignores")) test_autosuggestion_ignores();
     if (should_test_function("autosuggestion_combining")) test_autosuggestion_combining();
     if (should_test_function("autosuggest_suggest_special")) test_autosuggest_suggest_special();
+    if (should_test_function("wcstring_tok")) test_wcstring_tok();
     if (should_test_function("history")) history_tests_t::test_history();
     if (should_test_function("history_merge")) history_tests_t::test_history_merge();
     if (should_test_function("history_races")) history_tests_t::test_history_races();
diff --git a/tests/read.in b/tests/read.in
index f864287f3..7cf091129 100644
--- a/tests/read.in
+++ b/tests/read.in
@@ -1,3 +1,4 @@
+# vim: set filetype=fish:
 #
 # Test read builtin and IFS
 #
@@ -35,6 +36,8 @@ echo '' | read -l one two
 print_vars one two
 echo 'test' | read -l one two three
 print_vars one two three
+echo 'foo   bar   baz' | read -l one two three
+print_vars one two three
 
 echo
 set -l IFS
@@ -91,3 +94,25 @@ echo $foo
 echo $bar
 echo 'test' | read -n 1 foo
 echo $foo
+
+# read -0 tests
+
+echo
+echo '# read -z tests'
+echo -n 'testing' | read -lz foo
+echo $foo
+echo -n 'test ing' | read -lz foo
+echo $foo
+echo 'newline' | read -lz foo
+echo $foo
+echo -n 'test ing' | read -lz foo bar
+print_vars foo bar
+echo -ne 'test\0ing' | read -lz foo bar
+print_vars foo bar
+echo -ne 'foo\nbar' | read -lz foo bar
+print_vars foo bar
+echo -ne 'foo\nbar\0baz\nquux' | while read -lza foo
+    print_vars foo
+end
+
+true
diff --git a/tests/read.out b/tests/read.out
index 0de988314..d8cf948d1 100644
--- a/tests/read.out
+++ b/tests/read.out
@@ -16,6 +16,7 @@ two
 1 ''
 1 '' 1 ''
 1 'test' 1 '' 1 ''
+1 'foo' 1 'bar' 1 '  baz'
 
 1 'hello'
 1 'h' 1 'ello'
@@ -42,3 +43,14 @@ test
 tes
 tin
 t
+
+# read -z tests
+testing
+test ing
+newline
+
+1 'test' 1 'ing'
+1 'test' 1 ''
+1 'foo' 1 'bar'
+2 'foo' 'bar'
+2 'baz' 'quux'
diff --git a/wcstringutil.cpp b/wcstringutil.cpp
new file mode 100644
index 000000000..51ec1a1cf
--- /dev/null
+++ b/wcstringutil.cpp
@@ -0,0 +1,40 @@
+/** \file wcstringutil.cpp
+
+Helper functions for working with wcstring
+*/
+
+#include "config.h"
+
+#include "wcstringutil.h"
+
+typedef wcstring::size_type size_type;
+
+wcstring_range wcstring_tok(wcstring& str, const wcstring &needle, wcstring_range last)
+{
+    size_type pos = last.second == wcstring::npos ? wcstring::npos : last.first;
+    if (pos != wcstring::npos && last.second != wcstring::npos) pos += last.second;
+    if (pos != wcstring::npos && pos != 0) ++pos;
+    if (pos == wcstring::npos || pos >= str.size())
+    {
+        return std::make_pair(wcstring::npos, wcstring::npos);
+    }
+
+    if (needle.empty())
+    {
+        return std::make_pair(pos, wcstring::npos);
+    }
+
+    pos = str.find_first_not_of(needle, pos);
+    if (pos == wcstring::npos) return std::make_pair(wcstring::npos, wcstring::npos);
+
+    size_type next_pos = str.find_first_of(needle, pos);
+    if (next_pos == wcstring::npos)
+    {
+        return std::make_pair(pos, wcstring::npos);
+    }
+    else
+    {
+        str[next_pos] = L'\0';
+        return std::make_pair(pos, next_pos - pos);
+    }
+}
diff --git a/wcstringutil.h b/wcstringutil.h
new file mode 100644
index 000000000..73ca7ac69
--- /dev/null
+++ b/wcstringutil.h
@@ -0,0 +1,29 @@
+/** \file wcstringutil.h
+
+Helper functions for working with wcstring
+*/
+
+#ifndef FISH_WCSTRINGUTIL_H
+#define FISH_WCSTRINGUTIL_H
+
+#include <utility>
+#include "common.h"
+
+/**
+ typedef that represents a range in a wcstring.
+ The first element is the location, the second is the count.
+*/
+typedef std::pair<wcstring::size_type, wcstring::size_type> wcstring_range;
+
+/**
+   wcstring equivalent of wcstok(). Supports NUL.
+   For convenience and wcstok() compatibility, the first character of each
+   token separator is replaced with NUL.
+   Returns a pair of (pos, count).
+   Returns (npos, npos) when it's done.
+   Returns (pos, npos) when the token is already known to be the final token.
+   Note that the final token may not necessarily return (pos, npos).
+*/
+wcstring_range wcstring_tok(wcstring& str, const wcstring &needle, wcstring_range last = wcstring_range(0,0));
+
+#endif