all: split the delete_line_pattern into before and after

While at it, also add configuration for delete line before and after
for match-copyright section.

10 files changed, 301 insertions, 72 deletions

diff --git a/README.md b/README.md
index 5327e7e..0c48c70 100644
--- a/README.md
+++ b/README.md
@@ -109,12 +109,16 @@ prefix = "%"
 pattern = "^.*\\.(apk|app|bz2|csv|doc|docx|exe|gif|gz|jpeg|jpg|json|pdf|png|ppt|pptx|svg|svgz|tar|tgz|xls|xlsx|zip)$"
 
 [match-license]
-pattern = "^(//+|#+)\\s+(.*)governed by a BSD-style(.*)$"
+pattern = "^(//+|#+|/\\*+|<!--+)?\\s*(.*)governed by a BSD-style(.*)$"
 license_identifier = BSD-3-Clause
-delete_line_pattern = "^(//+|#+)\\s+license that(.*)$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
 
 [match-copyright]
-pattern = "^(//+|#+)\\s+Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<*(?<contact>.*)>.*$"
+pattern = "^(//+|#+|/\\*+|<!--+)?\\s*Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<(?<contact>.*)>.*$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
 ```
 
 The configuration use the `ini` file format.
@@ -194,9 +198,35 @@ regular expression.
 If there is a line that match with it, the value in
 "match-license::license_identifier" will replace the
 "default::license_identifier" value.
-If there is "delete_line_pattern" defined, it will search for line that match
-with that pattern and delete it.
-The "delete_line_pattern" can be defined zero or multiple times.
+
+If there is "delete_line_before" or "delete_line_after" defined, it will
+search for the pattern before and after the matched line and delete it.
+The "delete_line_before" and "delete_line_after" can be defined zero or
+multiple times.
+
+### match-copyright section
+
+The match-copyright section define the pattern to match with old copyright
+text.
+The regex must contains named group to capture copyright year, author, and
+contact.
+For example, given the following old copyright text,
+
+```
+Copyright 2022, John Doe <john.doe@email>. All rights reserved.
+```
+
+we can capture the year, author, and contact using the following regex,
+
+```
+^(//+|#+|/\\*+|<!--+)?\\s*Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<(?<contact>.*)>.*$"
+```
+
+The `match-copyright` section can also contains zero or more
+`delete_line_before` and `delete_line_after` pattern.
+The `delete_line_before` delete lines before matched line pattern, and
+`delete_line_after` contains regex to delete lines after matched line
+pattern.
 
 ## scan command
 
diff --git a/config.go b/config.go
index ed6ce10..dfa1f5f 100644
--- a/config.go
+++ b/config.go
@@ -56,10 +56,14 @@ pattern = "^.*\\.(apk|app|bz2|csv|doc|docx|exe|gif|gz|jpeg|jpg|json|pdf|png|ppt|
 [match-license]
 pattern = "^(//+|#+|/\\*+|<!--+)?\\s*(.*)governed by a BSD-style(.*)$"
 license_identifier = BSD-3-Clause
-delete_line_pattern = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
 
 [match-copyright]
 pattern = "^(//+|#+|/\\*+|<!--+)?\\s*Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<(?<contact>.*)>.*$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
 `
 
 type config struct {
diff --git a/file.go b/file.go
index 798790c..742290d 100644
--- a/file.go
+++ b/file.go
@@ -35,7 +35,8 @@ var reCopyrightText = regexp.MustCompile(`^(//+|#+|/\*+|<!--+)?\s?SPDX-FileCopyr
 // REUSE-IgnoreEnd
 
 type file struct {
-	matchLicense *matchLicense
+	matchCopyright *matchCopyright
+	matchLicense   *matchLicense
 
 	path string
 
@@ -226,7 +227,7 @@ func (f *file) apply(conv *SPDXConv) (err error) {
 		return err
 	}
 
-	f.applyDelete()
+	f.applyDelete(&conv.cfg)
 
 	// REUSE-IgnoreStart
 	line := fmt.Sprintf("%sSPDX-License-Identifier: %s%s",
@@ -360,19 +361,18 @@ func (f *file) applyCopyrightText(conv *SPDXConv) (err error) {
 		}
 		// Verify that the line actually match with one of
 		// match-copyright pattern.
-		var cmc *matchCopyright
-		for _, cmc = range conv.cfg.MatchCopyright {
-			if cmc.match(string(line)) {
-				f.copyrightYear = cmc.year
-				f.copyrightText = cmc.String()
+		for _, f.matchCopyright = range conv.cfg.MatchCopyright {
+			if f.matchCopyright.match(string(line)) {
+				f.copyrightYear = f.matchCopyright.year
+				f.copyrightText = f.matchCopyright.String()
 				break
 			}
 		}
-		if cmc == nil {
-			return fmt.Errorf(`%s: line %d does not match with any of match-license pattern`,
-				f.path, f.idxLicenseID)
+		if f.matchCopyright == nil {
+			return fmt.Errorf(`%s: line %d does not match with any of match-copyright pattern`,
+				f.path, f.idxCopyrightText)
 		}
-		if f.idxLicenseID >= 0 {
+		if f.idxCopyrightText >= 0 {
 			f.topLines[x] = nil
 		} else {
 			f.bottomLines[x] = nil
@@ -381,48 +381,131 @@ func (f *file) applyCopyrightText(conv *SPDXConv) (err error) {
 	return nil
 }
 
-func (f *file) applyDelete() {
-	var reDeleteLine []*regexp.Regexp
+func (f *file) applyDelete(cfg *config) {
+	var startat int
+	var lines [][]byte
 	if f.matchLicense != nil {
-		reDeleteLine = f.matchLicense.reDeleteLine
+		if f.idxLicenseID >= 0 {
+			startat = f.idxLicenseID
+			lines = f.topLines
+		} else {
+			startat = f.idxLicenseID + cfg.MaxLineMatch
+			lines = f.bottomLines
+		}
+		f.deleteLineBefore(f.matchLicense.reDeleteLineBefore, lines, startat-1)
+		f.deleteLineAfter(f.matchLicense.reDeleteLineAfter, lines, startat+1)
 	}
-	var lines [][]byte
-	for _, line := range f.topLines {
-		if line == nil {
-			continue
+	if f.matchCopyright != nil {
+		if f.idxCopyrightText >= 0 {
+			startat = f.idxCopyrightText
+			lines = f.topLines
+		} else {
+			startat = f.idxCopyrightText + cfg.MaxLineMatch
+			lines = f.bottomLines
 		}
-		var found bool
-		for _, redel := range reDeleteLine {
-			if redel.Match(line) {
-				found = true
-				break
-			}
+		f.deleteLineBefore(f.matchCopyright.reDeleteLineBefore, lines, startat-1)
+		f.deleteLineAfter(f.matchCopyright.reDeleteLineAfter, lines, startat+1)
+	}
+	f.trimDeletedLines()
+}
+
+func (f *file) deleteLineBefore(listre []*regexp.Regexp, lines [][]byte, startat int) {
+	if len(listre) == 0 {
+		return
+	}
+	var idxre = 0
+	var re = listre[idxre]
+	var line []byte
+	for x := startat; x >= 0; x-- {
+		if lines[x] == nil {
+			return
 		}
-		if found {
-			continue
+		line = lines[x]
+		if !re.Match(line) {
+			// Once the regex does not match with line, return
+			// immediately.
+			return
+		}
+		lines[x] = nil
+		idxre++
+		if idxre == len(listre) {
+			break
 		}
-		lines = append(lines, line)
+		re = listre[idxre]
 	}
-	f.topLines = lines
+	// Continue deleting with the next lines.
+	for x := len(f.lines) - 1; x >= 0; x-- {
+		line = f.lines[x]
+		if !re.Match(line) {
+			return
+		}
+		f.lines[x] = nil
+		idxre++
+		if idxre == len(listre) {
+			break
+		}
+		re = listre[idxre]
+	}
+}
 
-	lines = nil
-	for _, line := range f.bottomLines {
-		if line == nil {
-			continue
+func (f *file) deleteLineAfter(listre []*regexp.Regexp, lines [][]byte, startat int) {
+	if len(listre) == 0 {
+		return
+	}
+	var idxre = 0
+	var re = listre[idxre]
+	var line []byte
+	for x := startat; x < len(lines); x++ {
+		if lines[x] == nil {
+			return
 		}
-		var found bool
-		for _, redel := range reDeleteLine {
-			if redel.Match(line) {
-				found = true
-				break
-			}
+		line = lines[x]
+		if !re.Match(line) {
+			// Once the regex does not match with line, return
+			// immediately.
+			return
+		}
+		lines[x] = nil
+		idxre++
+		if idxre == len(listre) {
+			break
 		}
-		if found {
+		re = listre[idxre]
+	}
+	// Continue deleting with the next lines.
+	for x := 0; x < len(f.lines); x++ {
+		line = f.lines[x]
+		if !re.Match(line) {
+			return
+		}
+		f.lines[x] = nil
+		idxre++
+		if idxre == len(listre) {
+			break
+		}
+		re = listre[idxre]
+	}
+}
+
+func (f *file) trimDeletedLines() {
+	var newlines [][]byte
+	var lines [][]byte
+	if f.idxLicenseID >= 0 {
+		lines = f.topLines
+	} else {
+		lines = f.bottomLines
+	}
+	for x := 0; x < len(lines); x++ {
+		if lines[x] == nil {
 			continue
 		}
-		lines = append(lines, line)
+		newlines = append(newlines, lines[x])
+	}
+	if f.idxLicenseID >= 0 {
+		f.topLines = newlines
+	} else {
+		f.bottomLines = newlines
 	}
-	f.bottomLines = lines
 }
 
 // insertEmptyLine insert empty line after SPDX identifiers or any comments after it.
diff --git a/match_copyright.go b/match_copyright.go
index aa35950..723c63f 100644
--- a/match_copyright.go
+++ b/match_copyright.go
@@ -11,12 +11,18 @@ import (
 type matchCopyright struct {
 	rePattern *regexp.Regexp
 
+	reDeleteLineBefore []*regexp.Regexp
+	reDeleteLineAfter  []*regexp.Regexp
+
 	// Pattern to be searched in file.
 	Pattern string `ini:"match-copyright::pattern"`
 
 	year    string
 	author  string
 	contact string
+
+	DeleteLineBefore []string `ini:"match-copyright::delete_line_before"`
+	DeleteLineAfter  []string `ini:"match-copyright::delete_line_after"`
 }
 
 func (cmc *matchCopyright) init() (err error) {
@@ -27,6 +33,22 @@ func (cmc *matchCopyright) init() (err error) {
 			return fmt.Errorf(`%s: pattern %q: %w`, logp, cmc.Pattern, err)
 		}
 	}
+	cmc.reDeleteLineBefore = make([]*regexp.Regexp, len(cmc.DeleteLineBefore))
+	for x, pattern := range cmc.DeleteLineBefore {
+		re, err := regexp.Compile(pattern)
+		if err != nil {
+			return fmt.Errorf(`%s: delete_line_before %q: %w`, logp, pattern, err)
+		}
+		cmc.reDeleteLineBefore[x] = re
+	}
+	cmc.reDeleteLineAfter = make([]*regexp.Regexp, len(cmc.DeleteLineAfter))
+	for x, pattern := range cmc.DeleteLineAfter {
+		re, err := regexp.Compile(pattern)
+		if err != nil {
+			return fmt.Errorf(`%s: delete_line_after %q: %w`, logp, pattern, err)
+		}
+		cmc.reDeleteLineAfter[x] = re
+	}
 	return nil
 }
 
diff --git a/match_license.go b/match_license.go
index 1d15d7f..a9abb99 100644
--- a/match_license.go
+++ b/match_license.go
@@ -9,8 +9,10 @@ import (
 )
 
 type matchLicense struct {
-	rePattern    *regexp.Regexp
-	reDeleteLine []*regexp.Regexp
+	rePattern *regexp.Regexp
+
+	reDeleteLineBefore []*regexp.Regexp
+	reDeleteLineAfter  []*regexp.Regexp
 
 	// Pattern to be searched in file.
 	Pattern string `ini:"match-license::pattern"`
@@ -19,11 +21,15 @@ type matchLicense struct {
 	// value if Pattern match.
 	LicenseIdentifier string `ini:"match-license::license_identifier"`
 
-	// DeleteLinePattern zero or more pattern that will be search after
-	// Pattern match line.
-	// A line that match with this pattern will be deleted.
-	// An empty line stop the search.
-	DeleteLinePattern []string `ini:"match-license::delete_line_pattern"`
+	// DeleteLineBefore zero or more pattern that will be search before
+	// the matched Pattern line, and deleted if its match.
+	// Each pattern is executed in order until it does not match.
+	DeleteLineBefore []string `ini:"match-license::delete_line_before"`
+
+	// DeleteLineAfter zero or more pattern that will be search after
+	// the matched Pattern line, and deleted if its match.
+	// Each pattern is executed in order until it does not match.
+	DeleteLineAfter []string `ini:"match-license::delete_line_after"`
 }
 
 func (cml *matchLicense) init() (err error) {
@@ -34,13 +40,21 @@ func (cml *matchLicense) init() (err error) {
 			return fmt.Errorf(`%s: pattern %q: %w`, logp, cml.Pattern, err)
 		}
 	}
-	cml.reDeleteLine = make([]*regexp.Regexp, len(cml.DeleteLinePattern))
-	for x, pattern := range cml.DeleteLinePattern {
+	cml.reDeleteLineBefore = make([]*regexp.Regexp, len(cml.DeleteLineBefore))
+	for x, pattern := range cml.DeleteLineBefore {
+		re, err := regexp.Compile(pattern)
+		if err != nil {
+			return fmt.Errorf(`%s: delete_line_before %q: %w`, logp, pattern, err)
+		}
+		cml.reDeleteLineBefore[x] = re
+	}
+	cml.reDeleteLineAfter = make([]*regexp.Regexp, len(cml.DeleteLineAfter))
+	for x, pattern := range cml.DeleteLineAfter {
 		re, err := regexp.Compile(pattern)
 		if err != nil {
-			return fmt.Errorf(`%s: delete_line_pattern %q: %w`, logp, pattern, err)
+			return fmt.Errorf(`%s: delete_line_after %q: %w`, logp, pattern, err)
 		}
-		cml.reDeleteLine[x] = re
+		cml.reDeleteLineAfter[x] = re
 	}
 	return nil
 }
diff --git a/spdxconv_test.go b/spdxconv_test.go
index 77929e8..11a2a72 100644
--- a/spdxconv_test.go
+++ b/spdxconv_test.go
@@ -66,12 +66,22 @@ func TestInit(t *testing.T) {
 		MatchLicense: []*matchLicense{{
 			Pattern:           `^(//+|#+|/\*+|<!--+)?\s*(.*)governed by a BSD-style(.*)$`,
 			LicenseIdentifier: `BSD-3-Clause`,
-			DeleteLinePattern: []string{
+			DeleteLineBefore: []string{
+				`^(//+|#+|/\*+|<!--+)$`,
+			},
+			DeleteLineAfter: []string{
 				`^(//+|#+|/\*+|<!--+)?\s*license that can(.*)$`,
+				`^(//+|#+|\*+/|--+>)$`,
 			},
 		}},
 		MatchCopyright: []*matchCopyright{{
 			Pattern: `^(//+|#+|/\*+|<!--+)?\s*Copyright\s+(?<year>\d{4}),?\s+(?<author>.*)\s+<(?<contact>.*)>.*$`,
+			DeleteLineBefore: []string{
+				`^(//+|#+|/\*+|<!--+)$`,
+			},
+			DeleteLineAfter: []string{
+				`^(//+|#+|\*+/|--+>)$`,
+			},
 		}},
 	}
 	for _, mfc := range exp.MatchFileComment {
@@ -79,13 +89,25 @@ func TestInit(t *testing.T) {
 	}
 	for _, ml := range exp.MatchLicense {
 		ml.rePattern = regexp.MustCompile(ml.Pattern)
-		for _, dlp := range ml.DeleteLinePattern {
+		for _, dlp := range ml.DeleteLineBefore {
+			re := regexp.MustCompile(dlp)
+			ml.reDeleteLineBefore = append(ml.reDeleteLineBefore, re)
+		}
+		for _, dlp := range ml.DeleteLineAfter {
 			re := regexp.MustCompile(dlp)
-			ml.reDeleteLine = append(ml.reDeleteLine, re)
+			ml.reDeleteLineAfter = append(ml.reDeleteLineAfter, re)
 		}
 	}
 	for _, mc := range exp.MatchCopyright {
 		mc.rePattern = regexp.MustCompile(mc.Pattern)
+		for _, dlp := range mc.DeleteLineBefore {
+			re := regexp.MustCompile(dlp)
+			mc.reDeleteLineBefore = append(mc.reDeleteLineBefore, re)
+		}
+		for _, dlp := range mc.DeleteLineAfter {
+			re := regexp.MustCompile(dlp)
+			mc.reDeleteLineAfter = append(mc.reDeleteLineAfter, re)
+		}
 	}
 	test.Assert(t, `Init: loadConfig`, exp, conv.cfg)
 }
@@ -111,7 +133,7 @@ func TestScan(t *testing.T) {
 //spdxconv:header:path,license_id,idx_license_id,year,copyright_id,idx_copyright_id
 //spdxconv:regular
 test.go,match,1,2022,match,0,// ,
-test.html,match,2,2022,match,1,<!-- ," -->"
+test.html,match,4,2022,match,1,<!-- ," -->"
 test.sh,match,1,2022,match,0,# ,
 //spdxconv:binary
 //spdxconv:unknown`
diff --git a/testdata/Apply_test.txt b/testdata/Apply_test.txt
index 123dc28..8107ee8 100644
--- a/testdata/Apply_test.txt
+++ b/testdata/Apply_test.txt
@@ -50,10 +50,12 @@ pattern = "^.*\\.(apk|app|bz2|csv|doc|docx|exe|gif|gz|jpeg|jpg|json|pdf|png|ppt|
 [match-license]
 pattern = "^(//+|#+|/\\*+|<!--+)?\\s*(.*)governed by a BSD-style(.*)$"
 license_identifier = BSD-3-Clause
-delete_line_pattern = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_after = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
 
 [match-copyright]
 pattern = "^(//+|#+|/\\*+|<!--+)?\\s*Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<(?<contact>.*)>.*$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
 
 >>> without_spdx_license_id.go
 1
@@ -176,9 +178,6 @@ license that can be found in the LICENSE file.
 <<< with.html
 <!-- SPDX-License-Identifier: BSD-3-Clause -->
 <!-- SPDX-FileCopyrightText: 2022 Shulhan <ms@kilabit.info> -->
-<!--
-
--->
 
 1
 2
diff --git a/testdata/loadConfig/config_exists/spdxconv.cfg b/testdata/loadConfig/config_exists/spdxconv.cfg
index ec401a3..1f7ff68 100644
--- a/testdata/loadConfig/config_exists/spdxconv.cfg
+++ b/testdata/loadConfig/config_exists/spdxconv.cfg
@@ -3,13 +3,53 @@
 
 [default]
 license_identifier = GPL-3.0-only
+copyright_year = 2026
 file_copyright_text = Author <contact@email.local>
 max_line_match = 10
 
+[match-file-comment]
+pattern = "^.*\\.(adoc|asciidoc|c|cc|cpp|cs|dart|go|h|hh|hpp|java|js|jsx|jsonc|kt|kts|php|rs|sass|scss|swift|ts|tsx)$"
+prefix = "//"
+
+[match-file-comment]
+pattern = "^.*\\.(aff|bash|csh|dockerfile|env|gitignore|hcl|ipynb|make|pl|pm|py|ps1|rb|sh|tf|yaml|yml|zsh)$"
+prefix = "#"
+
+[match-file-comment]
+pattern = "^.*\\.(css)$"
+prefix = "/*"
+suffix = "*/"
+
+[match-file-comment]
+pattern = "^.*\\.(fxml|htm|html|html5|kml|markdown|md|xml)$"
+prefix = "<!--"
+suffix = "-->"
+
+[match-file-comment]
+pattern = "^.*\\.(lua|sql)$"
+prefix = "--"
+
+[match-file-comment]
+pattern = "^.*\\.(rst)$"
+prefix = ".."
+
+[match-file-comment]
+pattern = "^.*\\.(tex)$"
+prefix = "%"
+
+# File name that match with this pattern will have the ".license" file
+# created.
+[match-file-comment]
+pattern = "^.*\\.(apk|app|bz2|csv|doc|docx|exe|gif|gz|jpeg|jpg|json|pdf|png|ppt|pptx|svg|svgz|tar|tgz|xls|xlsx|zip)$"
+
 [match-license]
-pattern = "^(//+|#+)\\s+(.*)governed by a BSD-style(.*)$"
+pattern = "^(//+|#+|/\\*+|<!--+)?\\s*(.*)governed by a BSD-style(.*)$"
 license_identifier = BSD-3-Clause
-delete_line_pattern = "^(//+|#+)\\s+license that(.*)$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
 
 [match-copyright]
-pattern = "^(//+|#+)\\s+Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<*(?<contact>.*)>.*$"
+pattern = "^(//+|#+|/\\*+|<!--+)?\\s*Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<(?<contact>.*)>.*$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
diff --git a/testdata/scan/spdxconv.cfg b/testdata/scan/spdxconv.cfg
index 3b137b6..1cdfefb 100644
--- a/testdata/scan/spdxconv.cfg
+++ b/testdata/scan/spdxconv.cfg
@@ -42,10 +42,19 @@ prefix = "%"
 [match-file-comment]
 pattern = "^.*\\.(apk|app|bz2|csv|doc|docx|exe|gif|gz|jpeg|jpg|json|pdf|png|ppt|pptx|svg|svgz|tar|tgz|xls|xlsx|zip)$"
 
+# File name that match with this pattern will have the ".license" file
+# created.
+[match-file-comment]
+pattern = "^.*\\.(apk|app|bz2|csv|doc|docx|exe|gif|gz|jpeg|jpg|json|pdf|png|ppt|pptx|svg|svgz|tar|tgz|xls|xlsx|zip)$"
+
 [match-license]
 pattern = "^(//+|#+|/\\*+|<!--+)?\\s*(.*)governed by a BSD-style(.*)$"
 license_identifier = BSD-3-Clause
-delete_line_pattern = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|/\\*+|<!--+)?\\s*license that can(.*)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
 
 [match-copyright]
 pattern = "^(//+|#+|/\\*+|<!--+)?\\s*Copyright\\s+(?<year>\\d{4}),?\\s+(?<author>.*)\\s+<(?<contact>.*)>.*$"
+delete_line_before = "^(//+|#+|/\\*+|<!--+)$"
+delete_line_after = "^(//+|#+|\\*+/|--+>)$"
diff --git a/testdata/scan/test.html b/testdata/scan/test.html
index da2491e..bcfc0c3 100644
--- a/testdata/scan/test.html
+++ b/testdata/scan/test.html
@@ -1,5 +1,11 @@
 <!--
 Copyright 2022, Shulhan <ms@kilabit.info>. All rights reserved.
+-->
+<!--
 Use of this source code is governed by a BSD-style
 license that can be found in the LICENSE file.
--->
\ No newline at end of file
+-->
+
+1
+2
+3