def save_hrvr_grps(doc_id, page_no, bbox_list, cell_list, font_dict, b_fc_grps_bbox, b_fc_grps_cells, font_chg_grp_cells, new_cells, sfc_signature_list, fc_signature_list, f_res_bbox_n, f_res_cells_n): fname = '%s_HRVR.sh' % (page_no) #shname = os.path.join(ipath, str(doc_id), ci_odir, fname) shname = os.path.join(opath, str(doc_id), "MDB", fname) d = {} d['bbox'] = bbox_list d['cell'] = cell_list d['font'] = font_dict d['bbox_unordered'] = b_fc_grps_bbox d['cell_unordered'] = b_fc_grps_cells d['fc_cells'] = new_cells d['fc_sig_dict_list'] = fc_signature_list d['sfc_cells'] = font_chg_grp_cells d['sfc_sig_dict_list'] = sfc_signature_list d['hrvr_cells'] = f_res_cells_n d['hrvr_bboxs'] = f_res_bbox_n #datastore.write_data_fname(shname, isdb, isenc, d) datastore.write_data_fname(shname, isdb, 0, d) return
def write_cell_info_dict(doc_id, page_no, cell_info_dict): cid_odir = cfgObj.get_config('PageAnalysis', 'cell_info_dict_odir') fname = '%s.sh' % (page_no) ofname = os.path.join(opath, str(doc_id), cid_odir, fname) d = {} d['cell_info_dict'] = cell_info_dict datastore.write_data_fname(ofname, isdb, isenc, d) return
def write_font_data_dict(self, doc_id, page_no, cell_info_dict): cid_odir = cfgObj.get_config('MOD_DIRNAME', 'fontdict') fname = '%s.sh' % (page_no) print "CID PATH : ", cid_odir, fname ofname = os.path.join(self.opath, str(doc_id), cid_odir, fname) d = {} d['font_dict'] = cell_info_dict datastore.write_data_fname(ofname, self.isdb, self.isenc, d)
def write_number_curr_result(self, doc_id, page_no, cell_info_dict): cid_odir = cfgObj.get_config('PageAnalysis', 'number_curr_result') fname = '%s.sh' % (page_no) ofname = os.path.join(self.opath, str(doc_id), cid_odir, fname) d = {} d['data'] = cell_info_dict datastore.write_data_fname(ofname, self.isdb, self.isenc, d) return
def save_indexing(doc_id, page_no, CellInfoDict, inkey, level): fname = '%s_%s_%s.sh' % (page_no, inkey, level) #shname = os.path.join(ipath, str(doc_id), ci_odir, fname) shname = os.path.join(opath, str(doc_id), "MDD", fname) d = {} d['data'] = CellInfoDict #datastore.write_data_fname(shname, isdb, isenc, d) datastore.write_data_fname(shname, isdb, 0, d) return
def save_font_rm(doc_id, page_no, CellInfoDict): fname = '%s.sh' % (page_no) #shname = os.path.join(ipath, str(doc_id), ci_odir, fname) shname = os.path.join(opath, str(doc_id), "MDB", fname) d = {} d['data'] = CellInfoDict #datastore.write_data_fname(shname, isdb, isenc, d) datastore.write_data_fname(shname, isdb, 0, d) return
def update_cell_info_dict(self, doc_id, page_no, CellInfoDict): ci_odir = cfgObj.get_config('PageAnalysis', 'cell_info_dict_odir') fname = '%s.sh' % (page_no) #shname = os.path.join(ipath, str(doc_id), ci_odir, fname) shname = os.path.join(self.opath, str(doc_id), ci_odir, fname) d = {} d['cell_info_dict'] = CellInfoDict datastore.write_data_fname(shname, self.isdb, self.isenc, d) return
def write_projected_rm_linll(doc_id, page_no, level, rm_lines): prm_odir = cfgObj.get_config('PageAnalysis', 'projectedrmlnill_odir') fpath = os.path.join(opath, str(doc_id), prm_odir, '') #os.system('mkdir -p %s' %fpath) datastore.make_dirs(fpath) filename = '%s_%s.sh' % (str(page_no), level) fname = os.path.join(fpath, filename) d = {} d['data'] = rm_lines[:] datastore.write_data_fname(fname, isdb, isenc, d) return
def write_ph_info_dict(doc_id, pno, ph_info_dict): ph_odir = cfgObj.get_config('PageAnalysis', 'ParametricResults_odir') fname = '%s.sh' % (pno) #ofname = os.path.join(ipath, str(doc_id), ph_odir, fname) ofname = os.path.join(opath, str(doc_id), ph_odir, fname) datastore.rmfile(ofname) d = {} d['data'] = ph_info_dict datastore.write_data_fname(ofname, isdb, isenc, d) return
def update_cell_info_dict_level(doc_id, page_no, CellInfoDict, level): ci_odir = cfgObj.get_config('PageAnalysis', 'cell_info_dict_odir') fname = '%s_%s.sh' % (page_no, level) #shname = os.path.join(ipath, str(doc_id), ci_odir, fname) shname = os.path.join(opath, str(doc_id), ci_odir, fname) d = {} d['cell_info_dict'] = CellInfoDict #datastore.write_data_fname(shname, isdb, isenc, d) datastore.write_data_fname(shname, isdb, 0, d) return
def write_cagr_shelve(doc_id, page_no, CellInfoDict, level): ci_odir = cfgObj.get_config('PageAnalysis', 'cagr_result_odir') fname = '%s_%s.sh' % (page_no, level) #shname = os.path.join(ipath, str(doc_id), ci_odir, fname) shname = os.path.join(opath, str(doc_id), ci_odir, fname) datastore.make_dirs(os.path.join(opath, str(doc_id), ci_odir)) d = {} d['data'] = CellInfoDict datastore.write_data_fname(shname, isdb, isenc, d) return
def update_cell_info_dict_level(self, doc_id, page_no, CellInfoDict, level): ci_odir = cfgObj.get_config('PageAnalysis', 'cell_info_dict_odir') fname = '%s_%s.sh' % (page_no, level) #shname = os.path.join(ipath, str(doc_id), ci_odir, fname) shname = os.path.join(self.opath, str(doc_id), ci_odir, fname) #print shname, 'sh name' #for k, vs in CellInfoDict.items(): # print k, vs d = {} d['cell_info_dict'] = CellInfoDict print "ISSS", self.isdb datastore.write_data_fname(shname, self.isdb, self.isenc, d) return