Fix invalid css selectors in htmlreport (#3989)

Some plugins of cppcheck may use error ids that are not valid css
selectors. When the javascript script of htmlreport tries to use these
ids as selectors to display/hide categories of errors, this produces an
error inside the user's browser. The solution is to transform the error
ids that will be used as selectors in order to remove the invalid
characters. This transformation doesn't apply to what is displayed,
so that from a reader point of view, nothing is changed.

Signed-off-by: Cilyan Olowen <gaknar@gmail.com>

htmlreport/cppcheck-htmlreport

@@ -8,6 +8,7 @@ import locale
 import operator
 import optparse
 import os
+import re
 import sys
 import subprocess
 
@@ -498,6 +499,22 @@ def tr_str(td_th, line, id, cwe, severity, message, author, author_mail, date, a
     return '<tr%s>%s</tr>' % (tr_attributes, ret)
 
 
+def to_css_selector(tag):
+    # https://www.w3.org/TR/CSS2/syndata.html#characters
+    # Note: for our usage, we don't consider escaped characters
+    invalid_css_chars = re.compile(r"[^-_a-zA-Z0-9\u00A0-\uFFFF]")
+    invalid_start = re.compile(r"^([0-9]|-[0-9]|--)")
+    # Replace forbidden characters by an hyphen
+    valid_chars = invalid_css_chars.sub("-", tag)
+    # Check that the start of the tag doesn't break the rules
+    start_invalid = invalid_start.match(valid_chars)
+    if start_invalid:
+        # otherwise, append a token to make it valid
+        valid_chars_valid_start = "cpp" + valid_chars
+    else:
+        valid_chars_valid_start = valid_chars
+    return valid_chars_valid_start
+
 class AnnotateCodeFormatter(HtmlFormatter):
     errors = []
 
@@ -821,7 +838,12 @@ if __name__ == '__main__':
         stat_fmt = "\n            <tr><td><input type=\"checkbox\" class=\"idToggle\" onclick=\"toggleDisplay(this)\" id=\"{}\" name=\"{}\" checked></td><td>{}</td><td>{}</td></tr>"
         for occurrences in reversed(range(cnt_min, cnt_max + 1)):
             for _id in [k for k, v in sorted(counter.items()) if v == occurrences]:
-                stat_html.append(stat_fmt.format(_id, _id, dict(counter.most_common())[_id], _id))
+                stat_html.append(stat_fmt.format(
+                    to_css_selector(_id),
+                    to_css_selector(_id),
+                    dict(counter.most_common())[_id],
+                    _id
+                ))
 
         output_file.write(HTML_HEAD % (options.title, '', options.title, ''))
         output_file.write(filter_bar(filter_enabled))
@@ -881,7 +903,7 @@ if __name__ == '__main__':
                     tr_str('td', line, error["id"], cwe_url, error["severity"], error["msg"],
                            git_blame_dict.get('author', 'Unknown'), git_blame_dict.get('author-mail', '---'),
                            git_blame_dict.get('author-time', '---'),
-                           tr_class=error["id"] + ' sev_' + error["severity"] + ' issue',
+                           tr_class=to_css_selector(error["id"]) + ' sev_' + error["severity"] + ' issue',
                            message_class=message_class,
                            add_author=add_author_information,
                            htmlfile=htmlfile))