Carlos-Francisco Méndez-Cruz

Transform soft file in XML file

...@@ -18,9 +18,11 @@ __author__ = 'CMendezC' ...@@ -18,9 +18,11 @@ __author__ = 'CMendezC'
18 18
19 # Execution: 19 # Execution:
20 # python soft-2-xml.py 20 # python soft-2-xml.py
21 -# --inputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\soft-data 21 +# --inputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\soft-data-additional
22 # --outputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\xml-data 22 # --outputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\xml-data
23 # python soft-2-xml.py --inputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\soft-data --outputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\xml-data 23 # python soft-2-xml.py --inputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\soft-data --outputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\xml-data
24 +# Additional files
25 +# python soft-2-xml.py --inputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\soft-data-additional --outputPath C:\Users\cmendezc\Documents\GENOMICAS\gitlab_automatic-extraction-growth-conditions\data-sets\xml-data
24 26
25 ########################################################### 27 ###########################################################
26 # MAIN PROGRAM # 28 # MAIN PROGRAM #
...@@ -45,13 +47,14 @@ if __name__ == "__main__": ...@@ -45,13 +47,14 @@ if __name__ == "__main__":
45 print("Path to place output files: " + str(options.outputPath)) 47 print("Path to place output files: " + str(options.outputPath))
46 48
47 # Walk directory to read files 49 # Walk directory to read files
50 + processedFiles = 0
48 for path, dirs, files in os.walk(options.inputPath): 51 for path, dirs, files in os.walk(options.inputPath):
49 for f in files: 52 for f in files:
50 - if f.endswith("_family.txt"): 53 + if f.endswith("_family.soft"):
51 print("Processing...{}/{}".format(options.inputPath, f)) 54 print("Processing...{}/{}".format(options.inputPath, f))
52 softText = '' 55 softText = ''
53 with open(os.path.join(options.inputPath, f), "r", encoding="utf-8", errors="replace") as iFile: 56 with open(os.path.join(options.inputPath, f), "r", encoding="utf-8", errors="replace") as iFile:
54 - with open(os.path.join(options.outputPath, f.replace(".txt", ".xml")), "w", 57 + with open(os.path.join(options.outputPath, f.replace(".soft", ".xml")), "w",
55 encoding="utf-8") as oFile: 58 encoding="utf-8") as oFile:
56 oFile.write("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n\n<gse xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\nxsi:noNamespaceSchemaLocation=\"esquema-gcs.xsd\">\n\n") 59 oFile.write("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n\n<gse xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\nxsi:noNamespaceSchemaLocation=\"esquema-gcs.xsd\">\n\n")
57 for line in iFile: 60 for line in iFile:
...@@ -62,4 +65,5 @@ if __name__ == "__main__": ...@@ -62,4 +65,5 @@ if __name__ == "__main__":
62 # line = line.replace("\'", "&apos;") 65 # line = line.replace("\'", "&apos;")
63 oFile.write(line) 66 oFile.write(line)
64 oFile.write("\n</gse>\n") 67 oFile.write("\n</gse>\n")
65 - 68 + processedFiles+=1
69 + print("Processed files: {}".format(processedFiles))
......