natalie/indefero
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Add a unit test for monotone's basicio parser and compiler.

Browse Source 
Also note a couple of quirks and be less strict (for now) when it
comes to hash parsing and stanza lines without values.
This commit is contained in:
Thomas Keller
2011-03-23 00:26:26 +01:00
parent f08b5c5e3f
commit 5b5705fe90
2 changed files with 212 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files

102
src/IDF/Scm/Monotone/BasicIO.php
View File

@@ -21,6 +21,8 @@
#
# ***** END LICENSE BLOCK ***** */
require_once 'IDF/Scm/Exception.php';
/**
* Utility class to parse and compile basic_io stanzas
*
@@ -31,6 +33,11 @@ class IDF_Scm_Monotone_BasicIO
/**
* Parses monotone's basic_io format
*
* Known quirks:
* - does not handle multi-values starting with a hash '[]' (no known output)
* - does not validate hashes (should be /[0-9a-f]{40}/i)
* - does not handle forbidden \0
*
* @param string $in
* @return array of arrays
*/
@@ -54,50 +61,56 @@ class IDF_Scm_Monotone_BasicIO
$stanzaLine['key'] .= $ch;
}
// symbol w/o a value list
if ($pos >= $length || $in[$pos] == "\n") break;
if ($in[$pos] == '[') {
// ensure we don't look at a symbol w/o a value list
if ($pos >= $length || $in[$pos] == "\n") {
unset($stanzaLine['values']);
++$pos; // opening square bracket
$stanzaLine['hash'] = substr($in, $pos, 40);
$pos += 40;
++$pos; // closing square bracket
}
else
{
unset($stanzaLine['hash']);
$valCount = 0;
// if hashs and plain values are encountered in the same
// value list, we add the hash values as simple values as well
while ($in[$pos] == '"' || $in[$pos] == '[') {
$isHashValue = $in[$pos] == '[';
++$pos; // opening quote / bracket
$stanzaLine['values'][$valCount] = '';
while ($pos < $length) {
$ch = $in[$pos]; $pr = $in[$pos-1];
if (($isHashValue && $ch == ']')
||(!$isHashValue && $ch == '"' && $pr != '\\'))
break;
}
else {
if ($in[$pos] == '[') {
unset($stanzaLine['values']);
++$pos; // opening square bracket
while ($pos < $length && $in[$pos] != ']') {
$stanzaLine['hash'] .= $in[$pos];
++$pos;
$stanzaLine['values'][$valCount] .= $ch;
}
++$pos; // closing quote
++$pos; // closing square bracket
}
else
{
unset($stanzaLine['hash']);
$valCount = 0;
// if hashs and plain values are encountered in the same
// value list, we add the hash values as simple values as well
while ($in[$pos] == '"' || $in[$pos] == '[') {
$isHashValue = $in[$pos] == '[';
++$pos; // opening quote / bracket
$stanzaLine['values'][$valCount] = '';
while ($pos < $length) {
$ch = $in[$pos]; $pr = $in[$pos-1];
if (($isHashValue && $ch == ']')
||(!$isHashValue && $ch == '"' && $pr != '\\'))
break;
++$pos;
$stanzaLine['values'][$valCount] .= $ch;
}
++$pos; // closing quote
if (!$isHashValue) {
$stanzaLine['values'][$valCount] = str_replace(
array("\\\\", "\\\""),
array("\\", "\""),
$stanzaLine['values'][$valCount]
);
}
if (!$isHashValue) {
$stanzaLine['values'][$valCount] = str_replace(
array("\\\\", "\\\""),
array("\\", "\""),
$stanzaLine['values'][$valCount]
);
}
if ($pos >= $length)
break;
if ($pos >= $length)
break;
if ($in[$pos] == ' ') {
++$pos; // space
++$valCount;
if ($in[$pos] == ' ') {
++$pos; // space
++$valCount;
}
}
}
}
@@ -114,6 +127,12 @@ class IDF_Scm_Monotone_BasicIO
/**
* Compiles monotone's basicio format
*
* Known quirks:
* - does not validate keys for /[a-z_]+/
* - does not validate hashes (should be /[0-9a-f]{40}/i)
* - does not support intermixed value / hash formats
* - does not handle forbidden \0
*
* @param array $in Array of arrays
* @return string
*/
@@ -129,7 +148,7 @@ class IDF_Scm_Monotone_BasicIO
$maxkeylength = 0;
foreach ((array)$stanza as $lx => $line) {
if (!array_key_exists('key', $line)) {
if (!array_key_exists('key', $line) || empty($line['key'])) {
throw new IDF_Scm_Exception(
'"key" not found in basicio stanza '.$sx.', line '.$lx
);
@@ -157,13 +176,6 @@ class IDF_Scm_Monotone_BasicIO
$value).'"';
}
}
else
{
throw new IDF_Scm_Exception(
'neither "hash" nor "values" found in basicio '.
'stanza '.$sx.', line '.$lx
);
}
$out .= "\n";
}