#!/usr/bin/env python3

import requests
import sys

def fetchDataKeys(apiUrl):
    try:
        response = requests.get(apiUrl)
        try:
            data = response.json()
        except ValueError:
            print("Failed to parse JSON response.", file=sys.stderr)
            return []

        if response.status_code != 200:
            errorMsg = data.get("error", f"HTTP {response.status_code}")
            print(f"API error: {errorMsg}", file=sys.stderr)
            return []

        ### ignore these metaData 'keys' at the top level of the JSON structure
        metaKeys = {
            "downloadTime", "downloadTimeStamp", "browser", "q", "elapsedTimeMs",
            "itemCount", "totalMatchCount", "availableAssemblies", "maxItemsLimit",
            "statusCode", "statusMessage", "error"
        }

        dataKeys = [key for key in data if key not in metaKeys]
        return dataKeys

    except requests.exceptions.RequestException as e:
        print(f"Request failed: {e}", file=sys.stderr)
        return []

#########################################################################
if __name__ == "__main__":
    maxItemsOut = 4
    if len(sys.argv) != 2:
        print(f"Usage: findGenome.py <searchTerm>\n", file=sys.stderr)
        print(f"Using the UCSC API 'findGenome' function to search for an assembly", file=sys.stderr)
        print(f"in the UCSC genome browser system.  maxItemsOutput is set to {maxItemsOut}", file=sys.stderr)
        print(f"to limit the result output to four answers.\nExample::", file=sys.stderr)
        print(f"\t./findGenome.py GCF_000001405.40", file=sys.stderr)
        print(f"will print the link to the genome browser:\n\thttps://genome.ucsc.edu/cgi-bin/hgTracks?db=hg38\n")
        print(f"Note the discussion for the 'findGenome' function in the help doc:", file=sys.stderr)
        print(f"\thttps://genome.ucsc.edu/goldenPath/help/api.html", file=sys.stderr)
        sys.exit(1)

    searchTerm = sys.argv[1]
    print(f"# given search term: '{searchTerm}'", file=sys.stderr)
 
    ### the search term given with the 'q=searchTerm' can be any word that
    ### may match something in the assembly identifiers or version strings.
    ### see also: the 'findGenome' endpoint discussion in:
    ###    https://genome.ucsc.edu/goldenPath/help/api.html
    url = f"https://api.genome.ucsc.edu/findGenome?browser=mustExist;maxItemsOutput={maxItemsOut};q={searchTerm}"
#   print(f"{url}")
    keys = fetchDataKeys(url)
    ### search term may match to multiple browsers.  The returned results have
    ###  an hierarchical 'order' built into them.  The most 'important' ones
    ###  will be first.  Use maxItemsOutput=1 to return only the single
    ###    top priority result.
    for key in keys:
        print(f"https://genome.ucsc.edu/cgi-bin/hgTracks?db={key}")