Ticket #269: convert-to-ascii.py

#!/usr/bin/python

import h5py
import numpy as np
import os
#from lxml import etree
import xml.dom.minidom
from datetime import  date

csvFileName = "filelist.csv"

currentFolder = os.getcwd()
folder2m = "proc-2m-complexmask"
folder6m = "proc-6m-complexmask"
outfolder = "converted"


def write_node(doc, parent, name, value, attr={}):
    """
    :param doc: document DOM
    :param parent: parent node
    :param name: tag of the element
    :param value: value of the child text node
    :param attr: attribute dictionary
    
    :return: True if something was appended, otherwise False
    """
    if value is not None:
        node = doc.createElement(name)
        node.appendChild(doc.createTextNode(str(value)))
        for item in attr:
            node.setAttribute(item, attr[item])
        parent.appendChild(node)
        return True
    return False

def writeASCII(q,iq,header,outfilename):
        #Write out ascii to disk
        with open(outfilename, 'wb') as f:
                f.write(header)
                np.savetxt(f,zip(q,iq))
        f.close()

def writeCANSAS(q,iq,title,run,comment,sdd,origfile,filedate,outfile):
        """
        Create an XML document to contain the data
        """
        doc = xml.dom.minidom.Document()
        main_node = doc.createElement("SASroot")
        main_node.setAttribute("version", '1.0')
        main_node.setAttribute("xmlns", "cansas1d/%s" % '1.0')
        main_node.setAttribute("xmlns:xsi",
                               "http://www.w3.org/2001/XMLSchema-instance")
        main_node.setAttribute("xsi:schemaLocation",
                               "cansas1d/%s http://www.cansas.org/formats/1.0/cansas1d.xsd" % '1.0')

        doc.appendChild(main_node)

        entry_node = doc.createElement("SASentry")
        main_node.appendChild(entry_node)

        write_node(doc, entry_node, "Title", title)
        # for item in datainfo.run:
        #     runname = {}
        #     if item in datainfo.run_name and \
        #     len(str(datainfo.run_name[item])) > 1:
        #         runname = {'name': datainfo.run_name[item]}
        write_node(doc, entry_node, "Run", run)

        # Data info
        node = doc.createElement("SASdata")
        entry_node.appendChild(node)

        for i in range(len(q)):
                pt = doc.createElement("Idata")
                node.appendChild(pt)
                write_node(doc, pt, "Q", q[i], {'unit': '1/A'})
                write_node(doc, pt, "I", iq[i],{'unit': '1/cm'})

        # Sample info
        sample = doc.createElement("SASsample")
        entry_node.appendChild(sample)
        write_node(doc, sample, "ID", title)
        write_node(doc, sample, "details", title+" "+comment)

        # Instrument info
        instr = doc.createElement("SASinstrument")
        entry_node.appendChild(instr)
        write_node(doc, instr, "name", "I22 SAXS")

        #   Source
        source = doc.createElement("SASsource")
        source.setAttribute("name", "Diamond")
        instr.appendChild(source)

        write_node(doc, source, "radiation", "Synchrotron X-ray Source")

        #   Collimation
        coll = doc.createElement("SAScollimation")
        instr.appendChild(coll)

        #   Detectors
        det = doc.createElement("SASdetector")
        instr.appendChild(det)
        write_node(doc, det, "name", "Pilatus2M")

        # Processes info
        node = doc.createElement("SASprocess")
        entry_node.appendChild(node)

        write_node(doc, node, "name", "GDA/DAWN/Python")
        write_node(doc, node, "date", filedate)
        write_node(doc, node, "SASprocessnote", "Data originally converted to 1D with GDA/DAWN.\n canSAS XML generated from original nexus file :\n\t "+origfile+"\n by AJJ Script on "+date.today().isoformat()+"\n")

        #SASNote
        node = doc.createElement("SASnote")
        entry_node.appendChild(node)

        # Return the document, and the SASentry node associated with
        # the data we just wrote
        #return doc, entry_node

        # Write the file
        
        fd = open(outfile, 'w')
        fd.write(doc.toprettyxml())
        fd.close()





#Read in file list
inputFile = open(csvFileName,'r')
fileList = inputFile.read().splitlines()
inputFile.close()

#print fileList

#On each file, load HDF and extract Q, I(q)
for item in fileList:
        #print item
        (fileName,sample,pressure,description,sdd) = item.split(',')
        
        if (pressure == '' or sample == "Bad Run"):
                continue

        if sdd == '2':
                folder = folder2m
        elif sdd == '6':
                folder = folder6m
        else:
                continue
                
        fileNamePath = currentFolder+"/"+folder+"/"+fileName
        
        print "Processing ",folder+"/"+fileName
        
        outfilename = currentFolder+"/"+outfolder+"/"+fileName[12:18]
        filedate = "20"+fileName[33:35]+"-"+fileName[31:33]+"-"+fileName[29:31]
        #print outfilename
        
        dataFile = h5py.File(fileNamePath,"r")
        q = dataFile['/entry1/Pilatus2M_result/q'].value
        iq = dataFile['/entry1/Pilatus2M_result/data'][0,0,...]
        dataFile.close()
        
        header = "Title:"+sample+" at "+pressure+" bar\n"
        header += "Comment:"+description+"\n"
        header += "SDD:"+sdd+"\n"
        header += "Original reduced file:"+fileName+"\n"

        writeASCII(q,iq,header,outfilename+".txt")
        print "Done with ASCII!"
        writeCANSAS(q,iq,sample,fileName[12:18],description,sdd,fileName,filedate,outfilename+".xml")
        print "Done with canSAS!"