From fafe2f29158051eeff17c55374df4488b2344f7d Mon Sep 17 00:00:00 2001
From: Serge Titov <tit@westra.ru>
Date: Wed, 10 May 2023 13:37:03 +0600
Subject: [PATCH] Reduce excessive substring allocation, using preg_match's
 offset parameter (#595)

* use preg_match offset

* clean code

* fix codestyle

* preg_match returns false just on error
---
 src/Smalot/PdfParser/Element.php   |  2 +-
 src/Smalot/PdfParser/PDFObject.php | 69 ++++++++++++++++++++++++------
 2 files changed, 56 insertions(+), 15 deletions(-)

diff --git a/src/Smalot/PdfParser/Element.php b/src/Smalot/PdfParser/Element.php
index 4acfcf8b..bdd4afc3 100644
--- a/src/Smalot/PdfParser/Element.php
+++ b/src/Smalot/PdfParser/Element.php
@@ -107,7 +107,7 @@ public static function parse(string $content, ?Document $document = null, int &$
             $old_position = $position;
 
             if (!$only_values) {
-                if (!preg_match('/^\s*(?P<name>\/[A-Z0-9\._]+)(?P<value>.*)/si', substr($content, $position), $match)) {
+                if (!preg_match('/\G\s*(?P<name>\/[A-Z0-9\._]+)(?P<value>.*)/si', $content, $match, 0, $position)) {
                     break;
                 } else {
                     $name = ltrim($match['name'], '/');
diff --git a/src/Smalot/PdfParser/PDFObject.php b/src/Smalot/PdfParser/PDFObject.php
index 57c0f898..d038702d 100644
--- a/src/Smalot/PdfParser/PDFObject.php
+++ b/src/Smalot/PdfParser/PDFObject.php
@@ -592,18 +592,22 @@ public function getCommandsText(string $text_part, int &$offset = 0): array
                 case '/':
                     $type = $char;
                     if (preg_match(
-                        '/^\/([A-Z0-9\._,\+]+\s+[0-9.\-]+)\s+([A-Z]+)\s*/si',
-                        substr($text_part, $offset),
-                        $matches
+                        '/\G\/([A-Z0-9\._,\+]+\s+[0-9.\-]+)\s+([A-Z]+)\s*/si',
+                        $text_part,
+                        $matches,
+                        0,
+                        $offset
                     )
                     ) {
                         $operator = $matches[2];
                         $command = $matches[1];
                         $offset += \strlen($matches[0]);
                     } elseif (preg_match(
-                        '/^\/([A-Z0-9\._,\+]+)\s+([A-Z]+)\s*/si',
-                        substr($text_part, $offset),
-                        $matches
+                        '/\G\/([A-Z0-9\._,\+]+)\s+([A-Z]+)\s*/si',
+                        $text_part,
+                        $matches,
+                        0,
+                        $offset
                     )
                     ) {
                         $operator = $matches[2];
@@ -621,7 +625,14 @@ public function getCommandsText(string $text_part, int &$offset = 0): array
                         // get elements
                         $command = $this->getCommandsText($text_part, $offset);
 
-                        if (preg_match('/^\s*[A-Z]{1,2}\s*/si', substr($text_part, $offset), $matches)) {
+                        if (preg_match(
+                            '/\G\s*[A-Z]{1,2}\s*/si',
+                            $text_part,
+                            $matches,
+                            0,
+                            $offset
+                        )
+                        ) {
                             $operator = trim($matches[0]);
                             $offset += \strlen($matches[0]);
                         }
@@ -642,7 +653,14 @@ public function getCommandsText(string $text_part, int &$offset = 0): array
                         $offset = $strpos + 1;
                     }
 
-                    if (preg_match('/^\s*[A-Z]{1,2}\s*/si', substr($text_part, $offset), $matches)) {
+                    if (preg_match(
+                        '/\G\s*[A-Z]{1,2}\s*/si',
+                        $text_part,
+                        $matches,
+                        0,
+                        $offset
+                    )
+                    ) {
                         $operator = trim($matches[0]);
                         $offset += \strlen($matches[0]);
                     }
@@ -682,7 +700,14 @@ public function getCommandsText(string $text_part, int &$offset = 0): array
                         $command = substr($text_part, $offset, $strpos - $offset - 1);
                         $offset = $strpos;
 
-                        if (preg_match('/^\s*([A-Z\']{1,2})\s*/si', substr($text_part, $offset), $matches)) {
+                        if (preg_match(
+                            '/\G\s*([A-Z\']{1,2})\s*/si',
+                            $text_part,
+                            $matches,
+                            0,
+                            $offset
+                        )
+                        ) {
                             $operator = $matches[1];
                             $offset += \strlen($matches[0]);
                         }
@@ -693,19 +718,35 @@ public function getCommandsText(string $text_part, int &$offset = 0): array
                     if ('ET' == substr($text_part, $offset, 2)) {
                         break;
                     } elseif (preg_match(
-                        '/^\s*(?P<data>([0-9\.\-]+\s*?)+)\s+(?P<id>[A-Z]{1,3})\s*/si',
-                        substr($text_part, $offset),
-                        $matches
+                        '/\G\s*(?P<data>([0-9\.\-]+\s*?)+)\s+(?P<id>[A-Z]{1,3})\s*/si',
+                        $text_part,
+                        $matches,
+                        0,
+                        $offset
                     )
                     ) {
                         $operator = trim($matches['id']);
                         $command = trim($matches['data']);
                         $offset += \strlen($matches[0]);
-                    } elseif (preg_match('/^\s*([0-9\.\-]+\s*?)+\s*/si', substr($text_part, $offset), $matches)) {
+                    } elseif (preg_match(
+                        '/\G\s*([0-9\.\-]+\s*?)+\s*/si',
+                        $text_part,
+                        $matches,
+                        0,
+                        $offset
+                    )
+                    ) {
                         $type = 'n';
                         $command = trim($matches[0]);
                         $offset += \strlen($matches[0]);
-                    } elseif (preg_match('/^\s*([A-Z\*]+)\s*/si', substr($text_part, $offset), $matches)) {
+                    } elseif (preg_match(
+                        '/\G\s*([A-Z\*]+)\s*/si',
+                        $text_part,
+                        $matches,
+                        0,
+                        $offset
+                    )
+                    ) {
                         $type = '';
                         $operator = $matches[1];
                         $command = '';