Merge pull request #327 from FamilySearch/extract-labels

Extract labels into YAML

build/uri-def.py

@@ -61,6 +61,7 @@ def find_cat_tables(txt, g7, tagsets):
         sect = txt[i:j].replace('(Latter-Day Saint Ordinance)','`ord`') ## <- hack for ord-STAT
         for entry in re.finditer(r'`([A-Z0-9_]+)` *\| *(.*?) *[|\n]', sect):
             enum, meaning = entry.groups()
+            label = None
             h1 = sect.find('\n|',sect.rfind('\n#',0,entry.start()))+1
             h2 = sect.find('\n', h1)
             header = [_.strip() for _ in sect[h1:h2].strip('| ').split('|')]
@@ -92,6 +93,7 @@ def find_cat_tables(txt, g7, tagsets):
                 enums.setdefault(key,[]).append(pfx)
             elif 'month-' in pfx:
                 yamltype = 'month'
+                label = re.sub(r'(,| \().*', '', meaning[0])
                 k1 = sect.find('`', sect.find('URI for this calendar is', entry.start()))+1
                 k2 = sect.find('`', k1)
                 cal = sect[k1:k2]
@@ -109,15 +111,22 @@ def find_cat_tables(txt, g7, tagsets):
                 if pfx.startswith('g7:'):
                     if pfx[3:] in g7:
                         raise Exception(pfx+' defined as an enumeration and a '+g7[pfx[3:]][0])
-                    g7[pfx[3:]] = (yamltype, meaning)
+                    if label:
+                        g7[pfx[3:]] = (yamltype, meaning, None, label)
+                    else:
+                        g7[pfx[3:]] = (yamltype, meaning)
     return enums, calendars
 
 def find_calendars(txt, g7):
     """Looks for sections defining a `g7:cal-` URI"""
     for bit in re.finditer(r'#+ `[^`]*`[^\n]*\n+((?:\n+(?!#)|[^\n])*is `g7:(cal-[^`]*)`(?:\n+(?!#)|[^\n#])*)', txt):
         m = re.search('The epoch markers? ([`_A-Z0-9, and]+) (is|are) permitted', bit.group(1))
         marker = [] if not m else re.findall(r'[A-Z0-9_]+', m[1])
-        g7[bit.group(2)] = ('calendar',[bit.group(1)], marker)
+        m = re.match(r'^The ([A-Z][A-Za-z]* )+calendar', bit.group(1))
+        if m:
+            g7[bit.group(2)] = ('calendar',[bit.group(1)], marker, m.group(0)[4:-9])
+        else:
+            g7[bit.group(2)] = ('calendar',[bit.group(1)], marker)
 
 
 def joint_card(c1,c2):
@@ -215,7 +224,7 @@ def find_descriptions(txt, g7, ssp):
         if uri not in ssp:
             raise Exception('Found section for '+uri+' but no gedstruct')
         if uri.startswith('g7:'):
-            g7.setdefault(uri[3:],('structure',[],ssp[uri]))[1].extend((
+            g7.setdefault(uri[3:],('structure',[],ssp[uri],name.strip()))[1].extend((
                 name.strip(),
                 desc.strip()
             ))
@@ -352,6 +361,10 @@ def expand_prefix(txt, prefixes):
                 print('\nspecification:', file=fh)
                 for desc in g7[tag][1]:
                     print(yaml_str_helper('  - ', desc), file=fh)
+
+            if len(g7[tag]) > 3:
+                print('\nlabel:',repr(g7[tag][3]), file=fh)
+
             if g7[tag][0] == 'structure':
                 d = g7[tag][2]
                 payload = expand_prefix(d['pay'],prefixes) if d['pay'] is not None else 'null'

extracted-files/tags/ABBR

@@ -13,6 +13,8 @@ specification:
   - A short name of a title, description, or name used for sorting, filing, and
     retrieving records.
 
+label: 'Abbreviation'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/ADDR

@@ -36,6 +36,8 @@ specification:
     Duplicating information bloats files and introduces the potential for
     self-contradiction. ADR1, ADR2, and ADR3 should not be added to new files.
 
+label: 'Address'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures:

extracted-files/tags/ADOP

@@ -15,6 +15,8 @@ specification:
   - Creation of a legally approved child-parent relationship that does not
     exist biologically.
 
+label: 'Adoption'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/ADOP-FAMC

@@ -19,6 +19,8 @@ specification:
     performed the adoption; or by using a FAM where the adopting individual is
     the only HUSB/WIFE.
 
+label: 'Family child'
+
 payload: "@<https://gedcom.io/terms/v7/record-FAM>@"
 
 substructures:

extracted-files/tags/ADR1

@@ -17,6 +17,8 @@ specification:
 
     ADR1 should not be added to new files; see ADDRESS_STRUCTURE for more.
 
+label: 'Address Line 1'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/ADR2

@@ -17,6 +17,8 @@ specification:
 
     ADR2 should not be added to new files; see ADDRESS_STRUCTURE for more.
 
+label: 'Address Line 2'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/ADR3

@@ -17,6 +17,8 @@ specification:
 
     ADR3 should not be added to new files; see ADDRESS_STRUCTURE for more.
 
+label: 'Address Line 3'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/AGE

@@ -13,6 +13,8 @@ specification:
   - The age of the individual at the time an event occurred, or the age listed
     in the document.
 
+label: 'Age at event'
+
 payload: https://gedcom.io/terms/v7/type-Age
 
 substructures:

extracted-files/tags/AGNC

@@ -16,6 +16,8 @@ specification:
     or events, or an organization responsible for creating or archiving
     records.
 
+label: 'Responsible agency'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/ALIA

@@ -22,6 +22,8 @@ specification:
     mutually linked by symmetric pairs of ALIA pointers. A future version of
     this specification may adjust the definition of ALIA.
 
+label: 'Alias'
+
 payload: "@<https://gedcom.io/terms/v7/record-INDI>@"
 
 substructures:

extracted-files/tags/ANCI

@@ -13,6 +13,8 @@ specification:
   - Indicates an interest in additional research for ancestors of this
     individual. (See also DESI).
 
+label: 'Ancestor interest'
+
 payload: "@<https://gedcom.io/terms/v7/record-SUBM>@"
 
 substructures: {}

extracted-files/tags/ANUL

@@ -14,6 +14,8 @@ specification:
   - annulment
   - Declaring a marriage void from the beginning (never existed).
 
+label: 'Annulment'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/ASSO

@@ -31,6 +31,8 @@ specification:
         2 ASSO @I2@
         3 ROLE CLERGY
 
+label: 'Associates'
+
 payload: "@<https://gedcom.io/terms/v7/record-INDI>@"
 
 substructures:

extracted-files/tags/AUTH

@@ -15,6 +15,8 @@ specification:
     an unpublished source, this may be an individual, a government agency,
     church organization, or private organization.
 
+label: 'Author'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/BAPL

@@ -15,6 +15,8 @@ specification:
   - The event of baptism performed at age 8 or later by priesthood authority of
     The Church of Jesus Christ of Latter-day Saints. (See also BAPM)
 
+label: 'Baptism, Latter-Day Saint'
+
 payload: null
 
 substructures:

extracted-files/tags/BAPM

@@ -14,6 +14,8 @@ specification:
   - baptism
   - Baptism, performed in infancy or later. (See also BAPL and CHR.)
 
+label: 'Baptism'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/BARM

@@ -14,6 +14,8 @@ specification:
   - Bar Mitzvah
   - The ceremonial event held when a Jewish boy reaches age 13.
 
+label: 'Bar Mitzvah'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/BASM

@@ -15,6 +15,8 @@ specification:
   - The ceremonial event held when a Jewish girl reaches age 13, also known as
     “Bat Mitzvah.”
 
+label: 'Bas Mitzvah'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/BIRT

@@ -14,6 +14,8 @@ specification:
   - birth
   - Entering into life.
 
+label: 'Birth'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/BLES

@@ -15,6 +15,8 @@ specification:
   - Bestowing divine care or intercession. Sometimes given in connection with a
     naming ceremony.
 
+label: 'Blessing'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/BURI

@@ -14,6 +14,8 @@ specification:
   - burial
   - Disposing of the mortal remains of a deceased person.
 
+label: 'Burial'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/CALN

@@ -14,6 +14,8 @@ specification:
     from the holdings of a repository. Despite the word “number” in the name,
     may contain any character, not just digits.
 
+label: 'Call number'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures:

extracted-files/tags/CAST

@@ -16,6 +16,8 @@ specification:
     based on racial or religious differences, or differences in wealth,
     inherited rank, profession, or occupation.
 
+label: 'Caste'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures:

extracted-files/tags/CAUS

@@ -14,6 +14,8 @@ specification:
     death event to show cause of death, such as might be listed on a death
     certificate.
 
+label: 'Cause'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/CHAN

@@ -20,6 +20,8 @@ specification:
     recent, although only the most recent change is described by the DATE
     substructure.
 
+label: 'Change'
+
 payload: null
 
 substructures:

extracted-files/tags/CHIL

@@ -13,6 +13,8 @@ specification:
   - The child in a family, whether biological, adopted, foster, sealed, or
     other relationship.
 
+label: 'Child'
+
 payload: "@<https://gedcom.io/terms/v7/record-INDI>@"
 
 substructures:

extracted-files/tags/CHR

@@ -14,6 +14,8 @@ specification:
   - christening
   - Baptism or naming events for a child.
 
+label: 'Christening'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/CHRA

@@ -14,6 +14,8 @@ specification:
   - adult christening
   - Baptism or naming events for an adult person.
 
+label: 'Christening, adult'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/CITY

@@ -12,6 +12,8 @@ specification:
   - City
   - The name of the city used in the address. See ADDRESS_STRUCTURE for more.
 
+label: 'City'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/CONF

@@ -14,6 +14,8 @@ specification:
   - confirmation
   - Conferring full church membership.
 
+label: 'Confirmation'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/CONL

@@ -15,6 +15,8 @@ specification:
   - The religious event by which a person receives membership in The Church of
     Jesus Christ of Latter-day Saints. (See also CONF)
 
+label: 'Confirmation, Latter-Day Saint'
+
 payload: null
 
 substructures:

extracted-files/tags/CONT

@@ -14,6 +14,8 @@ specification:
     during serialization and is never present in parsed datasets. See Lines for
     more.
 
+label: 'Continued'
+
 payload: null
 
 substructures: {}

extracted-files/tags/COPR

@@ -13,6 +13,8 @@ specification:
   - A copyright statement, as appropriate for the copyright laws applicable to
     this data.
 
+label: 'Copyright'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures: {}

extracted-files/tags/CORP

@@ -13,6 +13,8 @@ specification:
   - The name of the business, corporation, or person that produced or
     commissioned the product.
 
+label: 'Corporate name'
+
 payload: http://www.w3.org/2001/XMLSchema#string
 
 substructures:

extracted-files/tags/CREA

@@ -16,6 +16,8 @@ specification:
     to the initial creation, it should not be modified after the structure is
     created.
 
+label: 'Creation'
+
 payload: null
 
 substructures:

extracted-files/tags/CREM

@@ -14,6 +14,8 @@ specification:
   - cremation
   - Disposal of the remains of a person’s body by fire.
 
+label: 'Cremation'
+
 payload: Y|<NULL>
 
 substructures:

extracted-files/tags/CROP

@@ -33,6 +33,8 @@ specification:
     -   TOP or TOP + HEIGHT exceed the image height.
     -   CROP applied to a non-image or image without a defined pixel unit.
 
+label: 'Crop'
+
 payload: null
 
 substructures: