Phase change data parsing

[IvanChernyshov]

Add functionality to extract phase change data as dataframes

Looks similar to #21 and #22, probably the same parser can be used with individual post-processors.

[AntoJASM]

Hello @IvanChernyshov,

I recently downloaded melting point data for organic molecules from the NIST database and successfully performed parsing. I identified all the molecules in question using re.search with a pattern that included capture groups. I did the same for some identifiers but they were not difficult to parse

My main concern (and what I focused on) was identifying the specific pattern that marks the beginning of a property within the downloaded text. For example, for melting point, it was "Tfus" in the request or the text obtained with Beautiful Soup.

I've developed a regex pattern specifically for Tfus entries that also handles ranges and extracts detailed metadata. Here's my pattern:

_PATTERN = r"""
    Tfus # Matches the literal string "Tfus"
    \s* # Matches any whitespace characters (spaces, tabs, newlines) zero or more times
    (?P<value>[0-9]+\.?[0-9]*) # Captures the primary numerical value (e.g., "368.1", "142.77")
                               # This group is named 'value'.
    \s* # Matches any whitespace characters zero or more times
    (?P<operator>(±|to)*) # Optionally captures an operator like "±" or "to". This group is named 'operator'.
                           # The '*' makes the entire operator group optional and allows for multiple occurrences if they somehow exist.
    \s* # Matches any whitespace characters zero or more times
    (?P<aux_value>[0-9]*\.*[0-9]*)? # Optionally captures an auxiliary numerical value for ranges. This group is named 'aux_value'.
                                    # The '?' makes this entire group optional.
    \s* # Matches any whitespace characters zero or more times
    (?P<unit>[A-Z]{1}) # Captures a single uppercase letter as a unit (e.g., "K"). This group is named 'unit'.
                       # IMPORTANT: If units can be multi-character like "kJ/mol", this needs adjustment, e.g., r"([\w/]+)"
    \s* # Matches any whitespace characters zero or more times
    (?P<method>(AVG)*N/A) # Captures method information like "AVG" or "N/A". This group is named 'method'.
                          # The '(AVG)*N/A' means zero or more "AVG" followed by "N/A".
                          # IMPORTANT: If it means "AVG" OR "N/A", change to (AVG|N/A)
    \s* # Matches any whitespace characters zero or more times
    (?P<ref>.*([0-9]{4}|values)) # Captures the reference text, ending with a 4-digit year or the Word "values". This group is named 'ref'.
    \s* # Matches any whitespace characters zero or more times
    (?P<comment>.*) # Captures any remaining text as a comment. This group is named 'comment'.
"""

I actually almost recreated the NIST tables for melting point, first I looked for lines with "Tfus" match and then I parsed them using the pattern that I shared you above.

This multi-line pattern is designed to precisely extract the value, unit, operator (for ranges or uncertainty), auxiliary value, method, reference, and comment from lines starting with "Tfus". I find named capture groups (e.g., (?P...)) particularly useful for clarity and direct DataFrame mapping.

I believe that creating a dedicated Python class to manage these parsing patterns for various NIST properties (like melting point, enthalpy of vaporization, etc.) would be highly beneficial. This class could encapsulate the regex logic, extract data from the raw text (potentially obtained via Beautiful Soup or similar), and then consolidate this extracted information into a pandas DataFrame. This approach would provide a extensible, and clean solution for extracting all properties from the NIST database.

I'd be happy to discuss this further if you find this approach interesting.

Looking forward to hearing your thoughts.