djlint 1.28.0 → 1.30.0

package/src/djlint/reformat.py

@@ -0,0 +1,63 @@
+"""Djlint reformat html files.
+
+Much code is borrowed from https://github.com/rareyman/HTMLBeautify, many thanks!
+"""
+
+import difflib
+from pathlib import Path
+
+from .formatter.compress import compress_html
+from .formatter.condense import clean_whitespace, condense_html
+from .formatter.css import format_css
+from .formatter.expand import expand_html
+from .formatter.indent import indent_html
+from .formatter.js import format_js
+from .settings import Config
+
+
+def formatter(config: Config, rawcode: str) -> str:
+    """Format a html string."""
+    if not rawcode:
+        return rawcode
+
+    # naturalize the line breaks
+    compressed = compress_html(("\n").join(rawcode.splitlines()), config)
+
+    expanded = expand_html(compressed, config)
+
+    condensed = clean_whitespace(expanded, config)
+
+    indented_code = indent_html(condensed, config)
+
+    beautified_code = condense_html(indented_code, config)
+
+    if config.format_css:
+        beautified_code = format_css(beautified_code, config)
+
+    if config.format_js:
+        beautified_code = format_js(beautified_code, config)
+
+    # preserve original line endings
+    line_ending = rawcode.find("\n")
+    if line_ending > -1 and rawcode[max(line_ending - 1, 0)] == "\r":
+        # convert \r?\n to \r\n
+        beautified_code = beautified_code.replace("\r", "").replace("\n", "\r\n")
+
+    return beautified_code
+
+
+def reformat_file(config: Config, this_file: Path) -> dict:
+    """Reformat html file."""
+    rawcode = this_file.read_bytes().decode("utf8")
+
+    beautified_code = formatter(config, rawcode)
+
+    if config.check is not True or config.stdin is True:
+        this_file.write_bytes(beautified_code.encode("utf8"))
+
+    out = {
+        str(this_file): list(
+            difflib.unified_diff(rawcode.splitlines(), beautified_code.splitlines())
+        )
+    }
+    return out

package/src/djlint/rules/H025.py

@@ -0,0 +1,65 @@
+"""rule H025: Check for orphans html tags."""
+import copy
+from typing import Any, Dict, List
+
+import regex as re
+
+from ..helpers import (
+    inside_ignored_linter_block,
+    inside_ignored_rule,
+    overlaps_ignored_block,
+)
+from ..lint import get_line
+from ..settings import Config
+
+
+def run(
+    rule: Dict[str, Any],
+    config: Config,
+    html: str,
+    filepath: str,
+    line_ends: List[Dict[str, int]],
+    *args: Any,
+    **kwargs: Any,
+) -> List[Dict[str, str]]:
+    """Check for orphans html tags."""
+    errors: List[Dict[str, str]] = []
+    open_tags: List[re.Match] = []
+    for match in re.finditer(
+        re.compile(
+            r"<(/?(\w+))\s*(" + config.attribute_pattern + r"|\s*)*\s*?>",
+            re.VERBOSE,
+        ),
+        html,
+    ):
+        if match.group(1) and not re.search(
+            re.compile(rf"^/?{config.always_self_closing_html_tags}\b", re.I | re.X),
+            match.group(1),
+        ):
+            # close tags should equal open tags
+            if match.group(1)[0] != "/":
+                open_tags.insert(0, match)
+            else:
+                for i, tag in enumerate(copy.deepcopy(open_tags)):
+                    if tag.group(2) == match.group(1)[1:]:
+                        open_tags.pop(i)
+                        break
+                else:
+                    # there was no open tag matching the close tag
+                    open_tags.insert(0, match)
+
+    for match in open_tags:
+        if (
+            overlaps_ignored_block(config, html, match) is False
+            and inside_ignored_rule(config, html, match, rule["name"]) is False
+            and inside_ignored_linter_block(config, html, match) is False
+        ):
+            errors.append(
+                {
+                    "code": rule["name"],
+                    "line": get_line(match.start(), line_ends),
+                    "match": match.group().strip()[:20],
+                    "message": rule["message"],
+                }
+            )
+    return errors

package/src/djlint/rules.yaml

@@ -0,0 +1,279 @@
+- rule:
+    name: T001
+    message: Variables should be wrapped in a single whitespace.
+    flags: re.DOTALL
+    exclude:
+    - handlebars
+    - golang
+    patterns:
+    # open
+    - '{{[^\s#/@^\-\+]+'
+    - '{%[\-\+][^\s]+'
+    - '{%[^\s|\-|\+]+'
+    # handlebars
+    - '[^{]{#[^\s-]+|^{#[^\s-]+'
+    - '[^{]{#-[^\s]+|^{#-[^\s]+'
+    - '[^{]{\/[^\s]+|^{\/[^\s]+'
+    - '[^{]{\@[^\s]+|^{\@[^\s]+'
+    # close
+    - '[^(\s|^|\-|\+)]+[}|%|#]}'
+    - '[^(\s|^)]+\-[}|%|#]}'
+    - '[^(\s|^)]+\+[}|%|#]}'
+    - '[^\s\-\+][ ]{2,}[}|%|#]}'
+    - '{[{|%|#]\-?\+?[ ]{2,}'
+- rule:
+    name: T002
+    message: Double quotes should be used in tags.
+    flags: re.DOTALL
+    patterns:
+    - "{%[ \t]*?(?:trans(?:late)?|with|extends|include|now)[\\s]+?(?:(?:(?!%}|').)+?=)?'(?:(?!%}|').)*?'(?:(?!%}).)*?%}"
+- rule:
+    name: T003
+    message: 'Endblock should have name. Ex: {% endblock body %}.'
+    flags: re.DOTALL
+    patterns:
+    - '{%\s*?endblock\s*?%}'
+- rule:
+    name: D004
+    message: (Django) Static urls should follow {% static path/to/file %} pattern.
+    flags: re.DOTALL
+    # this should be using the static path from django settings
+    patterns:
+    - <(?:link|img|script|source)\s[^\>]*?(?:href|src|srcset)=[\"\']/?static/?
+- rule:
+    name: J004
+    message: (Jinja) Static urls should follow {{ url_for('static'..) }} pattern.
+    flags: re.DOTALL
+    # this should be using the static path from django settings
+    patterns:
+    - <(?:link|img|script|source)\s[^\>]*?(?:href|src|srcset)=[\"\']/?static/?
+- rule:
+    name: H005
+    message: Html tag should have lang attribute.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <html\s*(?:(?!lang).)*>
+- rule:
+    name: H006
+    message: Img tag should have height and width attributes.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <img\b(?:(?!(?:height)=)[^>])*/?>
+    - <img\b(?:(?!(?:width)=)[^>])*/?>
+- rule:
+    name: H007
+    message: <!DOCTYPE ... > should be present before the html tag.
+    flags: re.DOTALL|re.I
+    patterns:
+    - ^<html
+- rule:
+    name: H008
+    message: Attributes should be double quoted.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <(?:\w+)\b(\"[^\"]*\"|'[^']*'|{[^}]*}|[^'\">{}])*(?:class|id|src|width|height|alt|style|lang|title|srcset|media)=\'[^\']*'
+- rule:
+    name: H009
+    message: Tag names should be lowercase.
+    flags: re.DOTALL
+    patterns:
+    - (?<=<)/?(?:HTML|BODY|DIV|P|SPAN|TABLE|TR|TD|TH|THEAD|TBODY|CODE|UL|OL|LI|H1|H2|H3|H4|H5|H6)
+- rule:
+    name: H010
+    message: Attribute names should be lowercase.
+    flags: re.DOTALL
+    patterns:
+    - <\w+[^\>]+?(?:CLASS|ID|SRC|WIDTH|HEIGHT|ALT|STYLE|LANG|TITLE|MEDIA|SRCSET)=
+- rule:
+    name: H011
+    message: Attribute values should be quoted.
+    flags: re.DOTALL|re.I|re.M|re.X
+    patterns:
+    - |
+      <(?:(?!meta)\w+)\b(\"[^\"]*\"|'[^']*'|{[^}]*}|[^'\">{}])*(?:class|id|src|width|height|alt|style|lang|title|href|action|method|checked|required|srcset)=[a-zA-Z_-]+
+    - <(?:meta)\s+?[^>]*?(?:class|id|src|alt|style|lang|title|href|action|method|name)=[a-zA-Z_-]+
+- rule:
+    name: H012
+    message: There should be no spaces around attribute =.
+    flags: re.DOTALL
+    patterns:
+    - <\w+?(\"[^\"]*\"|'[^']*'|{[^}]*}|[^'\">{}])*\s+=
+    - <\w+?(\"[^\"]*\"|'[^']*'|{[^}]*}|[^'\">{}])*=\s
+- rule:
+    name: H013
+    message: Img tag should have an alt attribute.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <img\b(?:(?!(?:alt)=)[^>])*/?>
+- rule:
+    name: H014
+    message: Found extra blank lines.
+    flags: re.DOTALL
+    patterns:
+    - "[^\n]{,10}\n{3,}"
+- rule:
+    name: H015
+    message: Follow h tags with a line break.
+    flags: re.DOTALL
+    patterns:
+    - </h\d?>(?:(?!(.+\r?\n){1,}).)*<[a-zA-Z]+\d?
+- rule:
+    name: H016
+    message: Missing title tag in html.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <html[^>]*?>(?:(?!<title>).)*</html>
+- rule:
+    name: H017
+    message: Void tags should be self closing.
+    flags: re.DOTALL|re.I
+    default: false
+    patterns:
+    - <(img|input|area|base|br|col[^(?:group)]|embed|hr|link|param|source|track|wbr|command|keygen|menuitem|path)(\b(\"[^\"]*\"|'[^']*'|{{[^}]*}}|{%[^%]*%}|{#[^#]*#}|[^'\">{}])*)(?<!/)>
+    - <(img|input|area|base|br|col|embed|hr|link|param|source|track|wbr|command|keygen|menuitem|path)>
+- rule:
+    name: D018
+    message: (Django) Internal links should use the {% url ... %} pattern.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <(?:a|div|span|input)\b[^>]*?\s(?:href|data-url|data-src|action)=[\"|'](?!(?:https?://)|javascript:|on\w+:|mailto:|tel:|data:)[\w|/]+
+    - <form\b[^>]*?(?:\saction)=[\"|'](?!(?:https?://)|javascript:|on\w+:|mailto:|tel:)[\w|/]+
+- rule:
+    name: J018
+    message: (Jinja) Internal links should use the {{ url_for() ... }} pattern.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <(?:a|div|span|input)\b[^>]*?\s(?:href|data-url|data-src|action)=[\"|'](?!(?:https?://)|javascript:|on\w+:|mailto:|tel:|data:)[\w|/]+
+    - <form\s+?[^>]*?(?:action)=[\"|'](?!(?:https?://)|javascript:|on\w+:|mailto:|tel:)[\w|/]+
+- rule:
+    name: H019
+    message: Replace 'javascript:abc()' with on_ event and real url.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <(?:a|div|span|input)\s+?[^>]*?(?:href|data-url)=[\"|']javascript:[\w|/]+
+    - <form\s+?[^>]*?(?:action)=[\"|']javascript:[\w|/]+
+- rule:
+    name: H020
+    message: Empty tag pair found. Consider removing.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <((?!td|li|th|dt|dd)\w+)\s*?>\s*?<\/\1>
+- rule:
+    name: H021
+    message: Inline styles should be avoided.
+    flags: re.I
+    patterns:
+    - <\w+\s(?:[^>]*\s)?style=(?=((?!>|{{|{%).)*>)
+- rule:
+    name: H022
+    message: Use HTTPS for external links.
+    flags: re.I
+    patterns:
+    - <\w+\s[^>]*?(?:href|data-url|action|src|url|srcset)=[\"|']http://[^>]*?>
+- rule:
+    name: H023
+    message: Do not use entity references.
+    flags: re.I
+    patterns:
+    - '&(?!(lt|gt|amp|quot|nbsp|ensp|emsp|thinsp))[#0-9a-z]{,30};'
+- rule:
+    name: H024
+    message: Omit type on scripts and styles.
+    flags: re.I
+    patterns:
+    - <(?:script|style)[^>]*?type=[\"|'](?:(?:text/css)|(?:text/javascript))[^>]*?>
+- rule:
+    name: H025
+    message: Tag seems to be an orphan.
+    python_module: djlint.rules.H025
+- rule:
+    name: H026
+    message: Empty id and class tags can be removed.
+    flags: re.I
+    patterns:
+    - <\w+\b[^(?:{(?:%|{|#))>]*?\b(class|id)\b=(\"\"|'')
+    - <\w+\b[^(?:{(?:%|{|#))>-]*?\b(class|id)\b[^=\"-]
+- rule:
+    name: T027
+    message: Unclosed string found in template syntax.
+    flags: re.I
+    patterns:
+    # for tags with 3/5/7... quotes
+    # for single quotes
+    - "{%((?:(?!'|%}).)*?(')(?:(?!\\2|%}).)*?\\2(?:(?!\\2|%}).)*?)*\\2(?:(?!\\2|%}).)*?%}"
+    # for double quotes
+    - '{%((?:(?!"|%}).)*?(")(?:(?!\2|%}).)*?\2(?:(?!\2|%}).)*?)*\2(?:(?!\2|%}).)*?%}'
+
+    # for single quotes
+    - "{{((?:(?!'|}}).)*?(')(?:(?!\\2|}}).)*?\\2(?:(?!\\2|}}).)*?)*\\2(?:(?!\\2|}}).)*?}}"
+    # for double quotes
+    - '{{((?:(?!"|}}).)*?(")(?:(?!\2|}}).)*?\2(?:(?!\2|}}).)*?)*\2(?:(?!\2|}}).)*?}}'
+
+    # for tags with a single quote
+    - "{%((?:(?!'|\"|%}).)*?('|\")(?:(?!\\2|%}).)*?)%}"
+    - "{{((?:(?!'|\"|}}).)*?('|\")(?:(?!\\2|}}).)*?)}}"
+
+- rule:
+    name: T028
+    message: Consider using spaceless tags inside attribute values. {%- if/for -%}
+    exclude:
+    - django
+    patterns:
+    - <(?:/?(?:\w+)\b(?:\"[^\"]*\"|'[^']*'|{[^}]*}|[^'\">{}/])*(?<!\bclass)=([\"'])(?:(?!\1).)*?({%)[^-])\s*?(?:if|for|else|end)
+    - <(?:/?(?:\w+)\b(?:\"[^\"]*\"|'[^']*'|{[^}]*}|[^'\">{}/])*(?<!\bclass)=([\"'])(?:(?!\1).)*?{%(?:(?!%}).)*(?:if|else|for|end)(?:(?!%}).)*[^-](%}))
+- rule:
+    name: H029
+    message: Consider using lowercase form method values.
+    patterns:
+    - <[fF][oO][rR][mM]\b(?:\"[^\"]*\"|'[^']*'|{[^}]*}|[^'\">{}/])*([mM][eE][tT][hH][oO][dD])=(([\"'])[a-zA-Z]*?[A-Z][a-zA-Z]*?\3)
+- rule:
+    name: H030
+    message: Consider adding a meta description.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <html[^>]*?>(?:(?!<meta[^>]*?name=([\"|'])description\b).)*</html>
+- rule:
+    name: H031
+    message: Consider adding meta keywords.
+    flags: re.DOTALL|re.I
+    patterns:
+    - <html[^>]*?>(?:(?!<meta[^>]*?name=([\"|'])keywords\b).)*</html>
+- rule:
+    name: T032
+    message: Extra whitespace found in template tags.
+    patterns:
+    - "{%(([\"|'](?:(?!'|\"|%}).)*?[\"|'])|[^(?:%}|'|\"|\n)])*?[ \t]{2,}"
+    - "{{(([\"|'](?:(?!'|\"|}}).)*?[\"|'])|[^(?:}}|'|\"|\n)])*?[ \t]{2,}"
+- rule:
+    name: H033
+    message: Extra whitespace found in form action.
+    patterns:
+    - <form[^>]+?action=['|"]\s
+    - <form[^>]+?action=(['|"])({{(?:(?!}}).)*}}|{%(?:(?!%}).)*%}|([^"'{]))*\s+?\1
+- rule:
+    name: T034
+    message: Did you intend to use {% ... %} instead of {% ... }%?
+    flags: re.DOTALL
+    patterns:
+    - '{%(?:(?!%}).)*}%'
+- rule:
+    name: H035
+    message: Meta tags should be self closing.
+    flags: re.DOTALL|re.I
+    default: false
+    patterns:
+    - <(meta)(\b(\"[^\"]*\"|'[^']*'|{{[^}]*}}|{%[^%]*%}|{#[^#]*#}|[^'\">{}])*)(?<!/)>
+    - <(meta)>
+- rule:
+    name: H036
+    message: Avoid use of <br> tags.
+    flags: re.I
+    default: false
+    patterns:
+    - <br\s*?\/?>
+- rule:
+    name: H037
+    message: Duplicate attribute found.
+    flags: re.I
+    patterns:
+    - <\w[^>]*?\s\K([a-z-:][a-z-]*?)(?==(?:\"[^\"]*\"|'[^']*'|{{(?:(?!}}).)*}}|{%(?:(?!%}).)*%}|{#(?:(?!#}).)*#}|[^'\">{}])*[^-:]\1=[^>]*?>)