troggle/parsers/drawings.py

import sys
import os
import types
import stat
import csv
import re
import datetime
from pathlib import Path

from functools import reduce

import settings
from troggle.core.models.survex import SingleScan, Wallet, DrawingFile
from troggle.core.models.troggle import DataIssue
from troggle.core.utils import save_carefully

'''Searches through all the :drawings: repository looking
for tunnel and therion files
'''

todo='''- Rename functions more consistently between tunnel and therion variants

- Recode rx_valid_ext to use profile suffix() function
'''

rx_valid_ext   = re.compile(r'(?i)\.(?:png|jpg|pdf|jpeg|gif|txt)$')

def find_dwg_file(dwgfile, path):
    '''Is given a line of text 'path' which may or may not contain a recognisable name of a scanned file
    which we have already seen when we imported all the files we could find in the surveyscans direstories
    '''
    wallet, scansfile = None, None
    mscansdir = re.search(r"(\d\d\d\d#X?\d+\w?|1995-96kh|92-94Surveybookkh|1991surveybook|smkhs)/(.*?(?:png|jpg|pdf|jpeg|gif|txt))$", path)
    if mscansdir:
        scanswalletl = Wallet.objects.filter(walletname=mscansdir.group(1))
        # This should be changed to properly detect if a list of folders is returned and do something sensible, not just pick the first.
        if len(scanswalletl):
            wallet = scanswalletl[0]
            if len(scanswalletl) > 1:
                message = f"! More than one scan FOLDER matches filter query. [{scansfilel[0]}]: {mscansdir.group(1)} {mscansdir.group(2)} {dwgfile.dwgpath} {path}"
                print(message)
                DataIssue.objects.create(parser='Tunnel', message=message)
            
        if wallet:
            scansfilel = wallet.singlescan_set.filter(name=mscansdir.group(2))
            if len(scansfilel):
                if len(scansfilel) > 1:
                    plist =[]
                    for sf in scansfilel:
                        plist.append(sf.ffile)
                    message = f"! More than one image FILENAME matches filter query. [{scansfilel[0]}]: {mscansdir.group(1)} {mscansdir.group(2)} {dwgfile.dwgpath} {path} {plist}"
                    print(message)
                    DataIssue.objects.create(parser='Tunnel', message=message)
                scansfile = scansfilel[0]

        if wallet:
            dwgfile.dwgwallets.add(wallet)
        if scansfile:
            dwgfile.scans.add(scansfile)
    
    elif path and not rx_valid_ext.search(path):
        name = os.path.split(path)[1]
        rdwgfilel = DrawingFile.objects.filter(dwgname=name)
        if len(rdwgfilel):
            if len(rdwgfilel) > 1:
                plist =[]
                for df in rdwgfilel:
                    plist.append(df.dwgname)
                message = f"! {len(rdwgfilel)} paths found with same name '{path}' {plist}"
                print(message)
                DataIssue.objects.create(parser='Tunnel', message=message, url=f'/dwgdataraw/{path}')
                rdwgfile = rdwgfilel[0]
                dwgfile.dwgcontains.add(rdwgfile)

    dwgfile.save()

def findwalletimage(therionfile, foundpath):
    '''Tries to link the drawing file (Therion format) to the referenced image (scan) file
    '''
    foundpath = foundpath.strip("{}")
    mscansdir = re.search(r"(\d\d\d\d#\d+\w?|1995-96kh|92-94Surveybookkh|1991surveybook|smkhs)", foundpath)
    if mscansdir:
        scanswalletl = Wallet.objects.filter(walletname=mscansdir.group(1))
        # This should be changed to properly detect if a list of folders is returned and do something sensible, not just pick the first.
        if len(scanswalletl):
            wallet = scanswalletl[0]
            if len(scanswalletl) > 1:
                message = "! More than one scan FOLDER matches filter query. [{}]: {} {} {}".format(therionfile, mscansdir.group(1), foundpath)
                print(message)
                DataIssue.objects.create(parser='Therion', message=message)
        if wallet:
            therionfile.dwgwallets.add(wallet)
            
            scanfilename = Path(foundpath).name
            scansfilel = wallet.singlescan_set.filter(name=scanfilename, wallet=wallet)
            if len(scansfilel):
                # message = f'! {len(scansfilel)} {scansfilel} = {scanfilename} found in the wallet specified {wallet.walletname}'
                # print(message)
                if len(scansfilel) > 1:
                    plist =[]
                    for sf in scansfilel:
                        plist.append(sf.ffile)
                    message = f"! More than one image FILENAME matches filter query. [{scansfilel[0]}]: {mscansdir.group(1)} {mscansdir.group(2)} {dwgfile.dwgpath} {path} {plist}"
                    print(message)
                    DataIssue.objects.create(parser='Therion', message=message)
                scansfile = scansfilel[0]
                therionfile.scans.add(scansfile)
            else:
                message = f'! Scanned file {scanfilename} mentioned in "{therionfile.dwgpath}" is not actually found in {wallet.walletname}'
                wurl = f'/survey_scans/{wallet.walletname}/'.replace("#",":")
                # print(message)
                DataIssue.objects.create(parser='Therion', message=message, url = wurl)


def findimportinsert(therionfile, imp):
    '''Tries to link the scrap (Therion format) to the referenced therion scrap
    '''
    pass

rx_xth_me = re.compile(r'xth_me_image_insert.*{.*}$', re.MULTILINE)
rx_scrap  = re.compile(r'^survey (\w*).*$', re.MULTILINE)
rx_input  = re.compile(r'^input (\w*).*$', re.MULTILINE)

def settherionfileinfo(filetuple):
    '''Read in the drawing file contents and sets values on the dwgfile object
    '''
    thtype, therionfile = filetuple
    
    ff = os.path.join(settings.DRAWINGS_DATA, therionfile.dwgpath)
    therionfile.filesize = os.stat(ff)[stat.ST_SIZE]
    if therionfile.filesize <= 0:
        message = f"! Zero length therion file {ff}"
        print(message)
        DataIssue.objects.create(parser='Therion', message=message, url=f'/dwgdataraw/{therionfile.dwgpath}')
        return
    fin = open(ff,'r')
    ttext = fin.read()
    fin.close()
    
    # The equivalent for a tunnel 'path' would be a .th2 'line wall'  or 'scrap'
    # print(len(re.findall(r"line", ttext)))
    if thtype=='th':
        therionfile.npaths = len(re.findall(r"^input ", ttext, re.MULTILINE))
    elif thtype=='th2':
        therionfile.npaths = len(re.findall(r"^line ", ttext, re.MULTILINE))
    therionfile.save()
  
    # scan and look for survex blocks that might have been included, and image scans (as for tunnel drawings)
    # which would populate dwgfile.survexfile
    
    # in .th2 files:
    # ##XTHERION## xth_me_image_insert {500 1 1.0} {1700 {}} ../../../expofiles/surveyscans/2014/01popped_elev1.jpeg 0 {}
    # scrap blownout -projection plan -scale [-81.0 -42.0 216.0 -42.0 0.0 0.0 7.5438 0.0 m]
    
    for xth_me in rx_xth_me.findall(ttext):
        # WORK IN PROGRESS. Do not clutter up the DataIssues list with this
        message = f'! Un-parsed image filename: {therionfile.dwgname} : {xth_me.split()[-3]} - {therionfile.dwgpath}'
        # print(message)
        # DataIssue.objects.create(parser='xTherion', message=message, url=f'/dwgdataraw/{therionfile.dwgpath}')
        # ! Un-parsed image filename: 107coldest : ../../../expofiles/surveyscans/2015/2015#20/notes.jpg - therion/plan/107coldest.th2

        with open('therionrefs.log', 'a') as lg:
            lg.write(message + '\n')

        findwalletimage(therionfile, xth_me.split()[-3])
        
    for inp in rx_input.findall(ttext):
        # if this 'input' is a .th2 file we have already seen, then we can assign this as a sub-file
        # but we would need to disentangle to get the current path properly
        message = f'! Un-set (?) Therion .th2 input: - {therionfile.dwgname} : {inp} - {therionfile.dwgpath}'
        #print(message)
        DataIssue.objects.create(parser='xTherion', message=message, url=f'/dwgdataraw/{therionfile.dwgpath}')
        findimportinsert(therionfile, inp)
    
    therionfile.save()
    
rx_skpath = re.compile(rb'<skpath')
rx_pcpath = re.compile(rb'<pcarea area_signal="frame".*?sfsketch="([^"]*)" sfstyle="([^"]*)"')

def settnlfileinfo(dwgfile):
    '''Read in the drawing file contents and sets values on the dwgfile object
    Should try to read the date too e.g. tunneldate="2010-08-16 22:51:57
    then we could display on the master calendar per expo.
    '''
    ff = os.path.join(settings.DRAWINGS_DATA, dwgfile.dwgpath)
    dwgfile.filesize = os.stat(ff)[stat.ST_SIZE]
    if dwgfile.filesize <= 0:
        message = f"! Zero length tunnel file {ff}"
        print(message)
        DataIssue.objects.create(parser='Tunnel', message=message, url=f'/dwgdataraw/{dwgfile.dwgpath}')
        return
    fin = open(ff,'rb')
    ttext = fin.read()
    fin.close()
   
    dwgfile.npaths = len(rx_skpath.findall(ttext))
    dwgfile.save()
    
    # example drawing file in Tunnel format.
    # <tunnelxml tunnelversion="version2009-06-21 Matienzo" tunnelproject="ireby" tunneluser="goatchurch" tunneldate="2009-06-29 23:22:17">
    # <pcarea area_signal="frame" sfscaledown="12.282584" sfrotatedeg="-90.76982" sfxtrans="11.676667377221136" sfytrans="-15.677173422877454" sfsketch="204description/scans/plan(38).png" sfstyle="" nodeconnzsetrelative="0.0">
    
    for path, style in rx_pcpath.findall(ttext):
        find_dwg_file(dwgfile, path.decode())
    
    # should also scan and look for survex blocks that might have been included, and image scans
    # which would populate dwgfile.survexfile

    dwgfile.save()

def setdrwfileinfo(dwgfile):
    '''Read in the drawing file contents and sets values on the dwgfile object,
    but these are SVGs, PDFs or .txt files, so there is no useful format to search for
    This function is a placeholder in case we thnk of a way to do something
    to recognise generic survex filenames.
    '''
    ff = Path(settings.DRAWINGS_DATA) / dwgfile.dwgpath
    dwgfile.filesize = ff.stat().st_size
    if dwgfile.filesize <= 0:
        message = f"! Zero length drawing file {ff}"
        print(message)
        DataIssue.objects.create(parser='drawings', message=message, url=f'/dwgdataraw/{dwgfile.dwgpath}')
        return

def load_drawings_files():
    '''Breadth first search of drawings directory looking for sub-directories and *.xml filesize
    
    Why do we have all this detection of file types/! Why not use get_mime_types ? 
    What is it all for ??
    
    We import JPG, PNG and SVG files; which have already been put on the server,
    but the upload form intentionally refuses to upload PNG and JPG (though it does allow SVG)
    '''
    all_xml = []
    drawdatadir = settings.DRAWINGS_DATA
    DrawingFile.objects.all().delete()
    DataIssue.objects.filter(parser='drawings').delete()
    DataIssue.objects.filter(parser='Therion').delete()
    DataIssue.objects.filter(parser='xTherion').delete()
    DataIssue.objects.filter(parser='Tunnel').delete()
    if(os.path.isfile('therionrefs.log')):
        os.remove('therionrefs.log')
    

    drawingsdirs = [ "" ]
    while drawingsdirs:
        drawdir = drawingsdirs.pop()
        for f in os.listdir(os.path.join(drawdatadir, drawdir)):
            if f[0] == "." or f[-1] == "~":
                continue
            lf = os.path.join(drawdir, f)
            ff = os.path.join(drawdatadir, lf)
            if os.path.isdir(ff):
                drawingsdirs.append(lf) # lunatic! adding to list in middle of list while loop!
            elif Path(f).suffix.lower() == ".txt":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('txt',dwgfile))
            elif Path(f).suffix.lower() == ".xml":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('xml',dwgfile))
            elif Path(f).suffix.lower() == ".th":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('th',dwgfile))
            elif Path(f).suffix.lower() == ".th2":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('th2',dwgfile))
            elif Path(f).suffix.lower() == ".pdf":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('pdf',dwgfile))
            elif Path(f).suffix.lower() == ".png":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('png',dwgfile))
            elif Path(f).suffix.lower() == ".svg":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('svg',dwgfile))
            elif Path(f).suffix.lower() == ".jpg":
                # Always creates new
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f[:-4])[1])
                dwgfile.save()
                all_xml.append(('jpg',dwgfile))
            elif Path(f).suffix == '': 
                # therion file
                dwgfile = DrawingFile(dwgpath=lf, dwgname=os.path.split(f)[1])
                dwgfile.save()
                all_xml.append(('',dwgfile))

    print(f' - {len(all_xml)} Drawings files found')

    for d in all_xml:
        if d[0] in ['pdf', 'txt', 'svg', 'jpg', 'png', '']:
            setdrwfileinfo(d[1])
        if d[0] == 'xml':
            settnlfileinfo(d[1])
        # important to import .th2 files before .th so that we can assign them when found in .th files
        if d[0] == 'th2':
            settherionfileinfo(d)
        if d[0] == 'th':
            settherionfileinfo(d)
           
    # for drawfile in DrawingFile.objects.all():
        # SetTunnelfileInfo(drawfile)