Stop spam: remove icpsr-metadata email address

acf/acf_parse-docx-to-xml.py

@@ -121,7 +121,7 @@ def write_error(details, error_msg):
 def prep_cell_text(raw_txt):
 
     #clean string--remove curly quotes and zero-width spaces
-    txt_step_1 = raw_txt.replace('’', "'").replace('“', '"').replace('”', '"').replace('\u200b', '').replace("\u2009", " ")
+    txt_step_1 = raw_txt.replace('’', "'").replace('“', '"').replace('”', '"').replace('\u200b', '').replace("\u2009", " ").replace('•', '')
 
     #split into list of strings, trim white space from each string
     txt_step_2 = [item.strip() for item in txt_step_1.splitlines()]
@@ -275,10 +275,14 @@ def parse_requirement_blocks(data, cell, temp_list, details, parent_position):
         current_position = f"{parent_position} - {state_code}"
 
         # Parse entities (as a list)
-        entities_match = entities_pattern.match(data_slice[1])
-        if entities_match:
-            entities_text = entities_match.group(1)
-            record['entities'] = [entity.strip() for entity in entities_text.split(";")]
+        try:
+            entities_match = entities_pattern.match(data_slice[1])
+            if entities_match:
+                entities_text = entities_match.group(1)
+                record['entities'] = [entity.strip() for entity in entities_text.split(";")]
+        except IndexError:
+            print(data_slice)
+            sys.exit(1)
 
         # Parse tags (as a list)
         tags_match = tags_pattern.match(data_slice[2])
@@ -314,16 +318,25 @@ def parse_to_dict(list_of_strings, cell, search_term):
 
             # if found, assign values to dictionary
             temp_dict = {
-                "current_position": t,
-                "number": t.split(' ', 1)[1].strip(),
+                "current_position": t.strip(),
                 "name": list_of_strings[t_idx+1],
                 "source": find_source_link(cell, list_of_strings[t_idx+1], t),
                 "start_idx": t_idx,
                 "end_idx": t_idx+1
             }
 
-            # we also need to get a form of the name to use as a dict key. Just do this for everything, but it's only used with Titles
-            temp_dict['title_key'] = f"{temp_dict["current_position"]}-{temp_dict["name"]}".replace(' ', '').lower()
+            # NOTE: we have at least one state that does not include Title #'s. Add handling for this...
+            try:
+                temp_dict['number'] = t.split(' ', 1)[1].strip()
+            except IndexError:
+                temp_dict['number'] = None
+
+            # we also need to get a form of the name to use as a dict key. Just do this for everything, but it's only used with Titles. We have to take additional steps in case this state doesn't include Title Numbers
+            if temp_dict['current_position'].lower() == search_term[0].lower():
+                temp_dict['current_position'] = f"{temp_dict['current_position']} {temp_dict['name']}"
+                temp_dict['title_key'] = temp_dict["current_position"]
+            else:
+                temp_dict['title_key'] = f"{temp_dict["current_position"]}-{temp_dict["name"]}".replace(' ', '').lower()
 
             #note that if this record uses multiple 'Part' names, we have to see if the altName (i.e., the second partName) is used
             if len(search_term) > 1:
@@ -571,7 +584,7 @@ def parse_tables(doc, details, record_data, object_type):
                             break
 
                     for index, item in enumerate(statutes):
-                        if any(item.startswith(phrase) for phrase in ['Requirements related to', 'Requirements for ', 'Regulations regarding ']):
+                        if any(item.startswith(phrase) for phrase in ['Requirements related ', 'Requirements for ', 'Regulations regarding ']):
                             req_index = index
                             if "Regulations" in statutes[index]:
                                 found_regs = True

acf/acf_parse_config.json

@@ -1,22 +1,23 @@
 {
-    "input_doc": "C:/testing/ACF/Texas/Texas_011723_CORrev - ag(MU).docx",
-    "out_dir": "C:/testing/ACF/Texas",
+    "input_doc": "C:/ACF/FINAL APPROVED STATE RECORDS/Nevada/Nevada_Copy.docx",
+    "out_dir": "C:/ACF/FINAL APPROVED STATE RECORDS/Florida",
     "xsd_file": "schema_final.xsd",
     "xsl_file": "statetemplatev5.xsl",
-    "category": "true",
+    "category": "false",
     "titleName": "Title",
-    "subtitleName": "Subtitle",
+    "subtitleName": "",
     "articleName": "Chapter",
-    "partName": ["Subchapter"],
+    "partName": [],
     "subPartName": "",
     "titleContent": "false",
-    "state": "Texas",
-    "state_code_pattern": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(TX\\s(?:Hum Res Code|Govt Code|Health & Safety Code|Fam Code)\\s\\u00A7\\s[\\d\\.]+)\\)?$",
-    "statute_pattern": "^(TX\\s(?:Hum Res Code|Govt Code|Health & Safety Code|Fam Code)\\s\u00A7\\s[\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+    "state": "Nevada",
+    "state_code_pattern": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(FLA\\.\\sSTAT\\.\\s\\u00A7\\s\\d+\\.\\d+)\\)?$",
+    "statute_pattern": "^(FLA\\.\\sSTAT\\.\\s\\u00A7\\s\\d+\\.\\d+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
     "patterns": {
         "base": "(.+)\\s?\\u2013\\s?(.*)\\s+",
         "Alabama": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(AL\\sCode\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
         "Arizona": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(AZ\\sRev\\sStat\\s\\u00A7\\s\\d+-[-\\w\\.]+)\\)?$",
+        "Colorado": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(CO\\sCode\\s\\u00A7\\s\\[-\\d\\.]+)\\)?$",
         "Connecticut": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(CT\\sGen\\sStat\\s\\u00A7\\s[-\\w]+)\\)?$",
         "Delaware": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+\\sDE\\sCode\\s\\u00A7\\s[-\\w]+)\\)?$",
         "Flandreau": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(Flandreau\\sSantee\\sSioux\\sTribal\\sCode\\s\\u00A7\\s[-\\d]+)\\)?$",
@@ -25,9 +26,11 @@
         "Illinois": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+\\sILCS\\s[-\\d\\.\\/]+)\\)?$",
         "Indiana": "(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(IN Code\\s\\u00A7\\s[-\\.\\w]+)\\)?$",
         "Kentucky": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(K\\.Y\\.\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
+        "Maine": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(M\\.R\\.S\\.\\s[-\\w]+\\s\\u00A7\\s[-\\w]+)\\)?$",
         "Maryland": "(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MD Code,?\\s(?:Family Law|Local Government|General Provisions|Health\\s*[-\\u2013]\\s*General|Human Services),?\\s\u00A7\\s\\d+(?:\\.\\d+)?-[-–\\.\\w]*)\\)?$",
         "Massachusetts": "(.+)\\s?\\u2013\\s?(.*)\\s+\\(?(MA Gen L.+\\d+[A-Za-z]*)\\s*\\)?$",
         "Michigan": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MI\\sComp\\sLaws\\s\\u00A7\\s[\\w\\.]+)\\)?$",
+        "Minnesota": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MN\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
         "Mississippi": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MS\\sCode\\s\\u00A7\\s?[-\\w\\.]+)\\)?$",
         "Navajo Nation": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+\\sN\\.N\\.C\\.\\s\\u00A7\\s\\d+),\\sp\\.\\d+\\)?$",
         "New Jersey": "(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(NJ Rev Stat \\u00A7 \\d+:\\d+[-\\.\\w]*)\\)?$",
@@ -41,13 +44,16 @@
         "South Carolina": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\((SC\\sCode\\s\\u00A7\\s[-\\d]+)\\)\\.?$",
         "Tennessee": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(TN Code\\s\\u00A7\\s[-\\.\\w]+)\\)?$",
         "Texas": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(TX\\s(?:Hum Res Code|Govt Code|Health & Safety Code|Fam Code)\\s\\u00A7\\s[\\d\\.]+)\\)?$",
+        "Vermont": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+[A-Z]?\\sV\\.S\\.A\\.\\s\\u00A7\\s[-\\w]+)\\)?$",
         "Virginia": "^(VA Code\\s\\u00A7\\s+\\d+(?:\\.\\d+)?-\\d+(?:\\.\\d+)?(?::\\d+)?)\\s+[\\u2013\\u2014]\\s+(.*)$",
+        "Wisconsin": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(WI\\sStat\\s\\u00A7\\s[\\d\\.]+)\\)?$",
         "Wyoming": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(WY\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
         "Yurok": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(YTC\\s[\\d\\.]+)\\)?$"
     },
     "state_statutes": {
         "Alabama": "^(AL\\sCode\\s\\u00A7\\s[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Arizona": "^(AZ\\sRev\\sStat\\s\\u00A7\\s\\d+-[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Colorado": "^(CO\\sCode\\s\\u00A7\\s\\[-\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Connecticut": "^(CT\\sGen\\sStat\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Delaware": "^(\\d+\\sDE\\sCode\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Flandreau": "^(Flandreau\\sSantee\\sSioux\\sTribal\\sCode\\s\\u00A7\\s[-\\d]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
@@ -56,9 +62,11 @@
         "Illinois": "^(\\d+\\sILCS\\s[-\\d\\.\\/]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Indiana": "^(IN Code\\s\\u00A7\\s[-\\.\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Kentucky": "^(K\\.Y\\.\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Maine": "^(M\\.R\\.S\\.\\s[-\\w]+\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Maryland": "^(MD Code,?\\s(?:Family Law|Local Government|General Provisions|Health\\s*[-\\u2013]\\s*General|Human Services),?\\s\u00A7\\s\\d+(?:\\.\\d+)?-[-\\.\\w]*)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Massachusetts": "^(MA Gen L.+\\d+[A-Za-z]*)\\s?[\\u2013\\u2014]\\s?(.*)",
         "Michigan": "^(MI\\sComp\\sLaws\\s\\u00A7\\s[\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Minnesota": "^(M\\.R\\.S\\.\\s[-\\w]+\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Mississippi": "^(MS\\sCode\\s\\u00A7\\s?[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Navajo Nation": "^(\\d+\\sN\\.N\\.C\\.\\s\\u00A7\\s\\d+),\\sp\\.\\d+\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "New Jersey": "^(NJ Rev Stat \\u00A7 \\d+:\\d+[-\\.\\w]*)\\s+[\u2013\u2014]\\s+(.*)$",
@@ -71,8 +79,10 @@
         "South Carolina": "^(SC\\sCode\\s\\u00A7\\s[-\\d]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Tennessee": "^(TN Code\\s\\u00A7\\s[-\\.\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Texas": "^(TX\\s(?:Hum Res Code|Govt Code|Health & Safety Code|Fam Code)\\s\u00A7\\s[\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Vermont": "^(\\d+[A-Z]?\\sV\\.S\\.A\\.\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Virginia": "(.+?)\\s?\\u2013\\s?(.*?)\\s+\\(?(VA Code\\s\\u00A7\\s+\\d+(?:\\.\\d+)?-\\d+(?:\\.\\d+)?(?::\\d+)?)\\)?$",
         "Wyoming": "^(WY\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Wisconsin": "^(WI\\sStat\\s\\u00A7\\s[\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Yurok": "^(YTC\\s[\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$"
     }  
 }

acf/state_configs/AZ_acf_parse_config.json

@@ -1,50 +1,86 @@
 {
-    "input_doc": "C:/testing/ACF/Arizona/Arizona_test.docx",
-    "out_dir": "C:/testing/ACF/Arizona",
+    "input_doc": "C:/ACF/FINAL APPROVED STATE RECORDS/Arizona/Arizona_Copy.docx",
+    "out_dir": "C:/ACF/FINAL APPROVED STATE RECORDS/Arizona",
     "xsd_file": "schema_final.xsd",
     "xsl_file": "statetemplatev5.xsl",
     "category": "false",
     "titleName": "Title",
     "subtitleName": "",
     "articleName": "Chapter",
-    "partName": "",
+    "partName": [],
     "subPartName": "",
     "titleContent": "false",
     "state": "Arizona",
     "state_code_pattern": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(AZ\\sRev\\sStat\\s\\u00A7\\s\\d+-[-\\w\\.]+)\\)?$",
     "statute_pattern": "^(AZ\\sRev\\sStat\\s\\u00A7\\s\\d+-[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
     "patterns": {
         "base": "(.+)\\s?\\u2013\\s?(.*)\\s+",
+        "Alabama": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(AL\\sCode\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
         "Arizona": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(AZ\\sRev\\sStat\\s\\u00A7\\s\\d+-[-\\w\\.]+)\\)?$",
         "Connecticut": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(CT\\sGen\\sStat\\s\\u00A7\\s[-\\w]+)\\)?$",
         "Delaware": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+\\sDE\\sCode\\s\\u00A7\\s[-\\w]+)\\)?$",
+        "Flandreau": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(Flandreau\\sSantee\\sSioux\\sTribal\\sCode\\s\\u00A7\\s[-\\d]+)\\)?$",
         "Florida": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(FLA\\.\\sSTAT\\.\\s\\u00A7\\s\\d+\\.\\d+)\\)?$",
         "Idaho": "(.+)\\s?\\u2013\\s?(.*)\\s+\\(?(I\\.C\\. Stat\\s.+\\d+[A-Za-z]*)\\)?$",
+        "Illinois": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+\\sILCS\\s[-\\d\\.\\/]+)\\)?$",
         "Indiana": "(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(IN Code\\s\\u00A7\\s[-\\.\\w]+)\\)?$",
+        "Kentucky": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(K\\.Y\\.\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
+        "Maine": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(M\\.R\\.S\\.\\s[-\\w]+\\s\\u00A7\\s[-\\w]+)\\)?$",
         "Maryland": "(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MD Code,?\\s(?:Family Law|Local Government|General Provisions|Health\\s*[-\\u2013]\\s*General|Human Services),?\\s\u00A7\\s\\d+(?:\\.\\d+)?-[-–\\.\\w]*)\\)?$",
         "Massachusetts": "(.+)\\s?\\u2013\\s?(.*)\\s+\\(?(MA Gen L.+\\d+[A-Za-z]*)\\s*\\)?$",
+        "Michigan": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MI\\sComp\\sLaws\\s\\u00A7\\s[\\w\\.]+)\\)?$",
+        "Minnesota": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MN\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
+        "Mississippi": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(MS\\sCode\\s\\u00A7\\s?[-\\w\\.]+)\\)?$",
+        "Navajo Nation": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+\\sN\\.N\\.C\\.\\s\\u00A7\\s\\d+),\\sp\\.\\d+\\)?$",
         "New Jersey": "(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(NJ Rev Stat \\u00A7 \\d+:\\d+[-\\.\\w]*)\\)?$",
+        "New York": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(NY\\s.+\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
         "Nevada": "(.+)\\s?\\u2013\\s?(.*)\\s+\\(?(NRS[\\s\\.][A-Za-z0-9\\.]+)\\)?$",
+        "North Dakota": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(N\\.D\\.\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
         "Ohio": "(.+?)\\s?\\u2013\\s?(.*?)\\s+\\(?(Ohio Rev Code\\s\\u00A7\\s+\\d+(?:\\.\\d*)?)\\)?$",
         "Oklahoma": "(.+?)\\s?\\u2013\\s?(.*?)\\s+\\(?(\\d+[A-Za-z]?\\sOK STAT \\u00A7\\s[\\d-]+([A-Za-z0-9\\-\\.]+)?)\\)?$",
         "Oregon": "(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(ORS\\s+[A-Za-z0-9\\.]+)\\)?$",
+        "Rhode Island": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(R\\.I\\.\\sStat\\s\\u00A7\\s[-\\d\\.]+)\\)?$",
+        "South Carolina": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\((SC\\sCode\\s\\u00A7\\s[-\\d]+)\\)\\.?$",
         "Tennessee": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(TN Code\\s\\u00A7\\s[-\\.\\w]+)\\)?$",
-        "Virginia": "^(VA Code\\s\\u00A7\\s+\\d+(?:\\.\\d+)?-\\d+(?:\\.\\d+)?(?::\\d+)?)\\s+[\\u2013\\u2014]\\s+(.*)$"
+        "Texas": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(TX\\s(?:Hum Res Code|Govt Code|Health & Safety Code|Fam Code)\\s\\u00A7\\s[\\d\\.]+)\\)?$",
+        "Vermont": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(\\d+[A-Z]?\\sV\\.S\\.A\\.\\s\\u00A7\\s[-\\w]+)\\)?$",
+        "Virginia": "^(VA Code\\s\\u00A7\\s+\\d+(?:\\.\\d+)?-\\d+(?:\\.\\d+)?(?::\\d+)?)\\s+[\\u2013\\u2014]\\s+(.*)$",
+        "Wisconsin": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(WI\\sStat\\s\\u00A7\\s[\\d\\.]+)\\)?$",
+        "Wyoming": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(WY\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\)?$",
+        "Yurok": "^(.+?)\\s?\\u2013\\s?(.*?)\\s?\\(?(YTC\\s[\\d\\.]+)\\)?$"
     },
     "state_statutes": {
+        "Alabama": "^(AL\\sCode\\s\\u00A7\\s[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Arizona": "^(AZ\\sRev\\sStat\\s\\u00A7\\s\\d+-[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Connecticut": "^(CT\\sGen\\sStat\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Delaware": "^(\\d+\\sDE\\sCode\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Flandreau": "^(Flandreau\\sSantee\\sSioux\\sTribal\\sCode\\s\\u00A7\\s[-\\d]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Florida": "^(FLA\\.\\sSTAT\\.\\s\\u00A7\\s\\d+\\.\\d+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Idaho": "^(I\\.C\\. Stat\\s.+\\d+[A-Za-z]*)\\s?[\\u2013\\u2014]\\s?(.*)",
+        "Illinois": "^(\\d+\\sILCS\\s[-\\d\\.\\/]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Indiana": "^(IN Code\\s\\u00A7\\s[-\\.\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Kentucky": "^(K\\.Y\\.\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Maine": "^(M\\.R\\.S\\.\\s[-\\w]+\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Maryland": "^(MD Code,?\\s(?:Family Law|Local Government|General Provisions|Health\\s*[-\\u2013]\\s*General|Human Services),?\\s\u00A7\\s\\d+(?:\\.\\d+)?-[-\\.\\w]*)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Massachusetts": "^(MA Gen L.+\\d+[A-Za-z]*)\\s?[\\u2013\\u2014]\\s?(.*)",
+        "Michigan": "^(MI\\sComp\\sLaws\\s\\u00A7\\s[\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Minnesota": "^(M\\.R\\.S\\.\\s[-\\w]+\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Mississippi": "^(MS\\sCode\\s\\u00A7\\s?[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Navajo Nation": "^(\\d+\\sN\\.N\\.C\\.\\s\\u00A7\\s\\d+),\\sp\\.\\d+\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "New Jersey": "^(NJ Rev Stat \\u00A7 \\d+:\\d+[-\\.\\w]*)\\s+[\u2013\u2014]\\s+(.*)$",
+        "New York": "^(NY\\s.+\\s\\u00A7\\s[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "North Dakota": "^(N\\.D\\.\\sStat\\s\\u00A7\\s[-\\w\\.\\s\\(\\)]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Oklahoma": "^(\\d+[A-Za-z]?\\sOK STAT \\u00A7\\s[\\d-]+[A-Za-z0-9\\-\\.]+)\\s?[\\u2013\\u2014]\\s?(.*)",
         "Ohio": "^(Ohio Rev Code\\s\\u00A7\\s+\\d+(?:\\.\\d*-)?)\\s+[\u2013\u2014]\\s+(.*)",
         "Oregon": "^(ORS\\s+[A-Za-z0-9\\.]+)\\s+[\\u2013\\u2014]\\s+(.*)$",
+        "Rhode Island": "^(R\\.I\\.\\sStat\\s\\u00A7\\s[-\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "South Carolina": "^(SC\\sCode\\s\\u00A7\\s[-\\d]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
         "Tennessee": "^(TN Code\\s\\u00A7\\s[-\\.\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
-        "Virginia": "(.+?)\\s?\\u2013\\s?(.*?)\\s+\\(?(VA Code\\s\\u00A7\\s+\\d+(?:\\.\\d+)?-\\d+(?:\\.\\d+)?(?::\\d+)?)\\)?$"
+        "Texas": "^(TX\\s(?:Hum Res Code|Govt Code|Health & Safety Code|Fam Code)\\s\u00A7\\s[\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Vermont": "^(\\d+[A-Z]?\\sV\\.S\\.A\\.\\s\\u00A7\\s[-\\w]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Virginia": "(.+?)\\s?\\u2013\\s?(.*?)\\s+\\(?(VA Code\\s\\u00A7\\s+\\d+(?:\\.\\d+)?-\\d+(?:\\.\\d+)?(?::\\d+)?)\\)?$",
+        "Wyoming": "^(WY\\sStat\\s\\u00A7\\s[-\\w\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Wisconsin": "^(WI\\sStat\\s\\u00A7\\s[\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$",
+        "Yurok": "^(YTC\\s[\\d\\.]+)\\s+[-\\u2013\\u2014]\\s+(.*)$"
     }  
 }