Split PHP into two lexers - PHP and PHTML.

The former is pure PHP code while the latter is PHP code in <? ?> tags, within HTML. Fixes #210.
2025-07-15 01:14:21 +02:00 · 2020-06-30 20:56:49 +10:00
parent 11501493c9
commit 2b9ea60d89
10 changed files with 229 additions and 118 deletions
--- a/.golangci.yml
+++ b/.golangci.yml
@ -20,6 +20,11 @@ linters:
    - wsl
    - gomnd
    - gocognit
+    - goerr113
+    - nolintlint
+    - testpackage
+    - godot
+    - nestif

 linters-settings:
  govet:
--- a/.travis.yml
+++ b/.travis.yml
@ -4,7 +4,7 @@ go:
  - "1.13.x"
 script:
  - go test -v ./...
-  - curl -sfL https://install.goreleaser.com/github.com/golangci/golangci-lint.sh | bash -s v1.22.2
+  - curl -sfL https://install.goreleaser.com/github.com/golangci/golangci-lint.sh | bash -s 1.26.0
  - ./bin/golangci-lint run
  - git clean -fdx .
 after_success:
--- a/lexers/circular/php.go
+++ b/lexers/circular/php.go
@ -1,15 +1,12 @@
 package circular

 import (
-	"strings"
-
 	. "github.com/alecthomas/chroma" // nolint
-	"github.com/alecthomas/chroma/lexers/h"
 	"github.com/alecthomas/chroma/lexers/internal"
 )

-// PHP lexer.
-var PHP = internal.Register(DelegatingLexer(h.HTML, MustNewLexer(
+// PHP lexer for pure PHP code (not embedded in HTML).
+var PHP = internal.Register(MustNewLexer(
 	&Config{
 		Name:            "PHP",
 		Aliases:         []string{"php", "php3", "php4", "php5"},
@ -19,12 +16,10 @@ var PHP = internal.Register(DelegatingLexer(h.HTML, MustNewLexer(
 		CaseInsensitive: true,
 		EnsureNL:        true,
 	},
-	Rules{
-		"root": {
-			{`<\?(php)?`, CommentPreproc, Push("php")},
-			{`[^<]+`, Other, nil},
-			{`<`, Other, nil},
-		},
+	phpCommonRules.Rename("php", "root"),
+))
+
+var phpCommonRules = Rules{
 	"php": {
 		{`\?>`, CommentPreproc, Pop(1)},
 		{`(<<<)([\'"]?)((?:[\\_a-z]|[^\x00-\x7f])(?:[\\\w]|[^\x00-\x7f])*)(\2\n.*?\n\s*)(\3)(;?)(\n)`, ByGroups(LiteralString, LiteralString, LiteralStringDelimiter, LiteralString, LiteralStringDelimiter, Punctuation, Text), nil},
@ -82,10 +77,4 @@ var PHP = internal.Register(DelegatingLexer(h.HTML, MustNewLexer(
 		{`(\$\{)(\S+)(\})`, ByGroups(LiteralStringInterpol, NameVariable, LiteralStringInterpol), nil},
 		{`[${\\]`, LiteralStringDouble, nil},
 	},
-	},
-).SetAnalyser(func(text string) float32 {
-	if strings.Contains(text, "<?php") {
-		return 0.5
-	}
-	return 0.0
-})))
+}
--- a/lexers/circular/phtml.go
+++ b/lexers/circular/phtml.go
@ -0,0 +1,34 @@
+package circular
+
+import (
+	"strings"
+
+	. "github.com/alecthomas/chroma" // nolint
+	"github.com/alecthomas/chroma/lexers/h"
+	"github.com/alecthomas/chroma/lexers/internal"
+)
+
+// PHTML lexer is PHP in HTML.
+var PHTML = internal.Register(DelegatingLexer(h.HTML, MustNewLexer(
+	&Config{
+		Name:            "PHTML",
+		Aliases:         []string{"phtml"},
+		Filenames:       []string{"*.phtml"},
+		MimeTypes:       []string{"application/x-php", "application/x-httpd-php", "application/x-httpd-php3", "application/x-httpd-php4", "application/x-httpd-php5"},
+		DotAll:          true,
+		CaseInsensitive: true,
+		EnsureNL:        true,
+	},
+	Rules{
+		"root": {
+			{`<\?(php)?`, CommentPreproc, Push("php")},
+			{`[^<]+`, Other, nil},
+			{`<`, Other, nil},
+		},
+	}.Merge(phpCommonRules),
+).SetAnalyser(func(text string) float32 {
+	if strings.Contains(text, "<?php") {
+		return 0.5
+	}
+	return 0.0
+})))
--- a/lexers/lexers_test.go
+++ b/lexers/lexers_test.go
@ -73,7 +73,7 @@ func TestLexers(t *testing.T) {
 			if os.Getenv("RECORD") == "true" {
 				// Update the expected file with the generated output of this lexer
 				f, err := os.Create(expectedFilename)
-				defer f.Close()
+				defer f.Close() // nolint: gosec
 				assert.NoError(t, err)
 				assert.NoError(t, formatters.JSON.Format(f, nil, chroma.Literator(actual...)))
 			} else {
--- a/lexers/testdata/php.actual
+++ b/lexers/testdata/php.actual
@ -1,10 +1,3 @@
-<!DOCTYPE html>
-<html>
-<body>
-
-<h1>My first PHP page</h1>
-<?php
-
 $docs = $modx->getIterator('modResource', ["parent" => 84]);

 foreach($docs as $doc){
@ -14,6 +7,3 @@ foreach($docs as $doc){
    // $doc->save();
 }
 // some comment
-?>
-</body>
-</html>
--- a/lexers/testdata/php.expected
+++ b/lexers/testdata/php.expected
@ -1,24 +1,4 @@
 [
-  {"type":"CommentPreproc","value":"\u003c!DOCTYPE html\u003e"},
-  {"type":"Text","value":"\n"},
-  {"type":"Punctuation","value":"\u003c"},
-  {"type":"NameTag","value":"html"},
-  {"type":"Punctuation","value":"\u003e"},
-  {"type":"Text","value":"\n"},
-  {"type":"Punctuation","value":"\u003c"},
-  {"type":"NameTag","value":"body"},
-  {"type":"Punctuation","value":"\u003e"},
-  {"type":"Text","value":"\n\n"},
-  {"type":"Punctuation","value":"\u003c"},
-  {"type":"NameTag","value":"h1"},
-  {"type":"Punctuation","value":"\u003e"},
-  {"type":"Text","value":"My first PHP page"},
-  {"type":"Punctuation","value":"\u003c/"},
-  {"type":"NameTag","value":"h1"},
-  {"type":"Punctuation","value":"\u003e"},
-  {"type":"Text","value":"\n"},
-  {"type":"CommentPreproc","value":"\u003c?php"},
-  {"type":"Text","value":"\n\n"},
  {"type":"NameVariable","value":"$docs"},
  {"type":"Text","value":" "},
  {"type":"Operator","value":"="},
@ -81,16 +61,6 @@
  {"type":"Text","value":"\n    "},
  {"type":"CommentSingle","value":"// $doc-\u003esave();\n"},
  {"type":"Punctuation","value":"}"},
-  {"type":"Text","value":" \n"},
-  {"type":"CommentSingle","value":"// some comment\n"},
-  {"type":"CommentPreproc","value":"?\u003e"},
  {"type":"Text","value":"\n"},
-  {"type":"Punctuation","value":"\u003c/"},
-  {"type":"NameTag","value":"body"},
-  {"type":"Punctuation","value":"\u003e"},
-  {"type":"Text","value":"\n"},
-  {"type":"Punctuation","value":"\u003c/"},
-  {"type":"NameTag","value":"html"},
-  {"type":"Punctuation","value":"\u003e"},
-  {"type":"Text","value":"\n"}
+  {"type":"CommentSingle","value":"// some comment\n"}
 ]
--- a/lexers/testdata/phtml.actual
+++ b/lexers/testdata/phtml.actual
@ -0,0 +1,19 @@
+<!DOCTYPE html>
+<html>
+<body>
+
+<h1>My first PHP page</h1>
+<?php
+
+$docs = $modx->getIterator('modResource', ["parent" => 84]);
+
+foreach($docs as $doc){
+    $q=$doc->content;
+    $doc->set("content", preg_replace("/Some value/i", "Replacement", $q));
+    print_r($doc->content);
+    // $doc->save();
+} 
+// some comment
+?>
+</body>
+</html>
--- a/lexers/testdata/phtml.expected
+++ b/lexers/testdata/phtml.expected
@ -0,0 +1,96 @@
+[
+  {"type":"CommentPreproc","value":"\u003c!DOCTYPE html\u003e"},
+  {"type":"Text","value":"\n"},
+  {"type":"Punctuation","value":"\u003c"},
+  {"type":"NameTag","value":"html"},
+  {"type":"Punctuation","value":"\u003e"},
+  {"type":"Text","value":"\n"},
+  {"type":"Punctuation","value":"\u003c"},
+  {"type":"NameTag","value":"body"},
+  {"type":"Punctuation","value":"\u003e"},
+  {"type":"Text","value":"\n\n"},
+  {"type":"Punctuation","value":"\u003c"},
+  {"type":"NameTag","value":"h1"},
+  {"type":"Punctuation","value":"\u003e"},
+  {"type":"Text","value":"My first PHP page"},
+  {"type":"Punctuation","value":"\u003c/"},
+  {"type":"NameTag","value":"h1"},
+  {"type":"Punctuation","value":"\u003e"},
+  {"type":"Text","value":"\n"},
+  {"type":"CommentPreproc","value":"\u003c?php"},
+  {"type":"Text","value":"\n\n"},
+  {"type":"NameVariable","value":"$docs"},
+  {"type":"Text","value":" "},
+  {"type":"Operator","value":"="},
+  {"type":"Text","value":" "},
+  {"type":"NameVariable","value":"$modx"},
+  {"type":"Operator","value":"-\u003e"},
+  {"type":"NameAttribute","value":"getIterator"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralStringSingle","value":"'modResource'"},
+  {"type":"Punctuation","value":","},
+  {"type":"Text","value":" "},
+  {"type":"Punctuation","value":"["},
+  {"type":"LiteralStringDouble","value":"\"parent\""},
+  {"type":"Text","value":" "},
+  {"type":"Operator","value":"=\u003e"},
+  {"type":"Text","value":" "},
+  {"type":"LiteralNumberInteger","value":"84"},
+  {"type":"Punctuation","value":"]);"},
+  {"type":"Text","value":"\n\n"},
+  {"type":"Keyword","value":"foreach"},
+  {"type":"Punctuation","value":"("},
+  {"type":"NameVariable","value":"$docs"},
+  {"type":"Text","value":" "},
+  {"type":"Keyword","value":"as"},
+  {"type":"Text","value":" "},
+  {"type":"NameVariable","value":"$doc"},
+  {"type":"Punctuation","value":"){"},
+  {"type":"Text","value":"\n    "},
+  {"type":"NameVariable","value":"$q"},
+  {"type":"Operator","value":"="},
+  {"type":"NameVariable","value":"$doc"},
+  {"type":"Operator","value":"-\u003e"},
+  {"type":"NameAttribute","value":"content"},
+  {"type":"Punctuation","value":";"},
+  {"type":"Text","value":"\n    "},
+  {"type":"NameVariable","value":"$doc"},
+  {"type":"Operator","value":"-\u003e"},
+  {"type":"NameAttribute","value":"set"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralStringDouble","value":"\"content\""},
+  {"type":"Punctuation","value":","},
+  {"type":"Text","value":" "},
+  {"type":"NameOther","value":"preg_replace"},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralStringDouble","value":"\"/Some value/i\""},
+  {"type":"Punctuation","value":","},
+  {"type":"Text","value":" "},
+  {"type":"LiteralStringDouble","value":"\"Replacement\""},
+  {"type":"Punctuation","value":","},
+  {"type":"Text","value":" "},
+  {"type":"NameVariable","value":"$q"},
+  {"type":"Punctuation","value":"));"},
+  {"type":"Text","value":"\n    "},
+  {"type":"NameOther","value":"print_r"},
+  {"type":"Punctuation","value":"("},
+  {"type":"NameVariable","value":"$doc"},
+  {"type":"Operator","value":"-\u003e"},
+  {"type":"NameAttribute","value":"content"},
+  {"type":"Punctuation","value":");"},
+  {"type":"Text","value":"\n    "},
+  {"type":"CommentSingle","value":"// $doc-\u003esave();\n"},
+  {"type":"Punctuation","value":"}"},
+  {"type":"Text","value":" \n"},
+  {"type":"CommentSingle","value":"// some comment\n"},
+  {"type":"CommentPreproc","value":"?\u003e"},
+  {"type":"Text","value":"\n"},
+  {"type":"Punctuation","value":"\u003c/"},
+  {"type":"NameTag","value":"body"},
+  {"type":"Punctuation","value":"\u003e"},
+  {"type":"Text","value":"\n"},
+  {"type":"Punctuation","value":"\u003c/"},
+  {"type":"NameTag","value":"html"},
+  {"type":"Punctuation","value":"\u003e"},
+  {"type":"Text","value":"\n"}
+]
--- a/regexp.go
+++ b/regexp.go
@ -160,6 +160,14 @@ func Tokenise(lexer Lexer, options *TokeniseOptions, text string) ([]Token, erro
 // Rules maps from state to a sequence of Rules.
 type Rules map[string][]Rule

+// Rename clones rules then a rule.
+func (r Rules) Rename(old, new string) Rules {
+	r = r.Clone()
+	r[new] = r[old]
+	delete(r, old)
+	return r
+}
+
 // Clone returns a clone of the Rules.
 func (r Rules) Clone() Rules {
 	out := map[string][]Rule{}