[Python] What is the better method to parse these data instead of regex+conditional statements?

Suppose you have a data set that looks like

name1: float

name2: float [unit]

name3: {str: float [unit], (...repeat)}

name4: datetime

name5: int

name6: alphanumeric

name7: strings (sentence/phrase)

name8: {{...can contain any of above but we ignore..}}

What is the appropriate parsing method from python? With regex I solve this partially but feels like I need additional hard coded if statements to get to a complete solution. I am looking for pointers or direction here. Something similar to pd.read_csv(...), perhaps with some pre-clean then throw it into a parser?

PATTERN_GENERIC = r'(?P<name>[^,:]*)(,|\s?(?P<content>.*)'
PATTERN_VALUE = r'(?P<name>[^:]*)[^0-9-]*(?P<content>[-+]?\d*\.\d+|\d+)'
header = {}
with open(f) as fopen:
for line in fopen:
# parse number only
results = re.search(PATTERN_VALUE,line.strip('\n'))
if results:
h = results.groupdict()
header[h['name'].replace(' ','')] = float(h['content'])
else: #treat as ID
results = re.search(PATTERN_GENERIC,line.strip('\n'))
if results:
h = results.groupdict()
header[h['name'].replace(' ','')] = h['content']


What I wish to achieve

col1 | col2

name | numbers

name | non-numbers

Currently I haven't write something to handle the {...} lines, I can do that with a specific if statement. Also, it falls short on whenever there is a number it is seen as a number which isn't true for alphanumeric case (they are IDs not values). To solve that I probably have to write a pattern to differentiate alphanumeric vs numbers.

Continue reading...