def handle_caps(element):
"""Handle small caps.
Replace '°xxx' by '<span class="sc">xxx</span>'.
"""
import re
pattern = r"([^°]*)°([^\s\.,)+/:]*)(.*)"
# Find text to display in small caps
result = re.match(pattern, element.attrib["val"].encode(ENCODING))
# Initialize loop variables
previous_span = None
index = 0
while result:
before = result.group(1).decode(ENCODING)
sc = result.group(2).decode(ENCODING)
after = result.group(3).decode(ENCODING)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "sc"
span.text = sc
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after.encode(ENCODING))
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
def handle_font(element):
"""Replace '{xxx}' by '<span class="ipa">xxx</span>'.
"""
import re
# Find text to display in IPA
pattern = r"([^{}]*){([^}]*)}(.*)"
result = re.match(pattern, element.attrib["val"])
# Initialize loop variables
previous_span = None
index = 0
while result:
before = result.group(1)
ipa = result.group(2)
after = result.group(3)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "ipa"
span.text = ipa
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after)
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
def handle_pinyin(element):
"""Replace '@xxx' by '<span class="pinyin">xxx</span>'.
"""
import re
# Find pinyin
pattern = r"([^@]*)@(\w*)(.*)"
result = re.match(pattern, element.attrib["val"])
# Initialize loop variables
previous_span = None
index = 0
while result:
before = result.group(1)
pinyin = result.group(2)
after = result.group(3)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "pinyin"
span.text = pinyin
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after)
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
def handle_caps(element):
"""Handle small caps.
Replace '°xxx' by '<span class="sc">xxx</span>'.
"""
import re
pattern = r"([^°]*)°([^\s\.,)+/:]*)(.*)"
# Find text to display in small caps
result = re.match(pattern, element.attrib["val"].encode(ENCODING))
# Initialize loop variables
previous_span = None
index = 0
while result:
before = result.group(1).decode(ENCODING)
sc = result.group(2).decode(ENCODING)
after = result.group(3).decode(ENCODING)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "sc"
span.text = sc
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after.encode(ENCODING))
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
def handle_pinyin(element):
"""Replace '@xxx' by '<span class="pinyin">xxx</span>'.
"""
import re
# Find pinyin
pattern = r"([^@]*)@(\w*)(.*)"
result = re.match(pattern, element.attrib["val"])
# Initialize loop variables
previous_span = None
index = 0
while result:
before = result.group(1)
pinyin = result.group(2)
after = result.group(3)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "pinyin"
span.text = pinyin
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after)
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
def handle_font(element):
"""Replace '{xxx}' by '<span class="ipa">xxx</span>'.
"""
import re
# Find text to display in IPA
pattern = r"([^{}]*){([^}]*)}(.*)"
result = re.match(pattern, element.attrib["val"])
# Initialize loop variables
previous_span = None
index = 0
while result:
before = result.group(1)
ipa = result.group(2)
after = result.group(3)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "ipa"
span.text = ipa
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after)
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
def handle_fv(element):
"""Replace 'fv:xxx' and '|fv{xxx}' by '<span class="vernacular">xxx</span>'.
"""
import re
# Find text to display in vernacular font
pattern = r"(([^:\|]*)fv:([^\s\.,)]*)(.*))|(([^:\|]*)\|fv{([^}]*)}(.*))"
result = re.match(pattern, element.attrib["val"])
# Initialize loop variables
previous_span = None
index = 0
while result:
if result.group(1) is not None:
before = result.group(2)
vernacular = result.group(3)
after = result.group(4)
elif result.group(5) is not None:
before = result.group(6)
vernacular = result.group(7)
after = result.group(8)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "vernacular"
span.text = vernacular
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after)
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
def handle_fv(element):
"""Replace 'fv:xxx' and '|fv{xxx}' by '<span class="vernacular">xxx</span>'.
"""
import re
# Find text to display in vernacular font
pattern = r"(([^:\|]*)fv:([^\s\.,)]*)(.*))|(([^:\|]*)\|fv{([^}]*)}(.*))"
result = re.match(pattern, element.attrib["val"])
# Initialize loop variables
previous_span = None
index = 0
while result:
if result.group(1) is not None:
before = result.group(2)
vernacular = result.group(3)
after = result.group(4)
elif result.group(5) is not None:
before = result.group(6)
vernacular = result.group(7)
after = result.group(8)
# Handle previous span or element
if previous_span is None:
element.text = before
else:
previous_span.tail = before
# Create span
span = Element("span")
span.attrib["class"] = "vernacular"
span.text = vernacular
# Insert span in element
element.insert(index, span)
# Update result
result = re.match(pattern, after)
if not result:
span.tail = after
# Update loop variables
previous_span = span
index += 1
return element
