tools/make-packet-dcm.py

   1 #!/usr/bin/env python3
   2 import os.path
   3 import sys
   4 import itertools
   5 import lxml.etree
   6
   7 # This utility scrapes the DICOM standard document in DocBook format, finds the appropriate tables,
   8 # and extracts the data needed to build the lists of DICOM attributes, UIDs and value representations.
   9
  10 # If the files part05.xml, part06.xml and part07.xml exist in the current directory, use them.
  11 # Otherwise, download the current release from the current DICOM official sources.
  12 if os.path.exists("part05.xml"):
  13     print("Using local part05 docbook.", file=sys.stderr)
  14     part05 = lxml.etree.parse("part05.xml")
  15 else:
  16     print("Downloading part05 docbook...", file=sys.stderr)
  17     part05 = lxml.etree.parse("http://dicom.nema.org/medical/dicom/current/source/docbook/part05/part05.xml")
  18 if os.path.exists("part06.xml"):
  19     print("Using local part06 docbook.", file=sys.stderr)
  20     part06 = lxml.etree.parse("part06.xml")
  21 else:
  22     print("Downloading part06 docbook...", file=sys.stderr)
  23     part06 = lxml.etree.parse("http://dicom.nema.org/medical/dicom/current/source/docbook/part06/part06.xml")
  24 if os.path.exists("part07.xml"):
  25     print("Using local part07 docbook.", file=sys.stderr)
  26     part07 = lxml.etree.parse("part07.xml")
  27 else:
  28     print("Downloading part07 docbook...", file=sys.stderr)
  29     part07 = lxml.etree.parse("http://dicom.nema.org/medical/dicom/current/source/docbook/part07/part07.xml")
  30 dbns = {'db':'http://docbook.org/ns/docbook', 'xml':'http://www.w3.org/XML/1998/namespace'}
  31
  32 # When displaying the dissected packets, some attributes are nice to include in the description of their parent.
  33 include_in_parent = {"Patient Position",
  34                      "ROI Number",
  35                      "ROI Name",
  36                      "Contour Geometric Type",
  37                      "Observation Number",
  38                      "ROI Observation Label",
  39                      "RT ROI Interpreted Type",
  40                      "Dose Reference Structure Type",
  41                      "Dose Reference Description",
  42                      "Dose Reference Type",
  43                      "Target Prescription Dose",
  44                      "Tolerance Table Label",
  45                      "Beam Limiting Device Position Tolerance",
  46                      "Number of Fractions Planned",
  47                      "Treatment Machine Name",
  48                      "RT Beam Limiting Device Type",
  49                      "Beam Number",
  50                      "Beam Name",
  51                      "Beam Type",
  52                      "Radiation Type",
  53                      "Wedge Type",
  54                      "Wedge ID",
  55                      "Wedge Angle",
  56                      "Material ID",
  57                      "Block Tray ID",
  58                      "Block Name",
  59                      "Applicator ID",
  60                      "Applicator Type",
  61                      "Control Point Index",
  62                      "Nominal Beam Energy",
  63                      "Cumulative Meterset Weight",
  64                      "Patient Setup Number"}
  65
  66 # Data elements are listed in three tables in Part 6:
  67 # * Table 6-1. Registry of DICOM Data Elements
  68 # * Table 7-1. Registry of DICOM File Meta Elements
  69 # * Table 8-1. Registry of DICOM Directory Structuring Elements
  70 # All three tables are in the same format and can be merged for processing.
  71
  72 # The Command data elements (used only in networking), are listed in two tables in Part 7:
  73 # * Table E.1-1. Command Fields
  74 # * Table E.2-1. Retired Command Fields
  75 # The Retired Command Fields are missing the last column. For processing here,
  76 # we just add a last column with "RET", and they can be parsed with the same
  77 # as for the Data elements.
  78
  79 data_element_tables=["table_6-1", "table_7-1", "table_8-1"]
  80
  81 def get_trs(document, table_id):
  82     return document.findall(f"//db:table[@xml:id='{table_id}']/db:tbody/db:tr",
  83                             namespaces=dbns)
  84
  85 data_trs = sum((get_trs(part06, table_id) for table_id in data_element_tables), [])
  86 cmd_trs = get_trs(part07, "table_E.1-1")
  87 retired_cmd_trs = get_trs(part07, "table_E.2-1")
  88
  89 def get_texts_in_row(tr):
  90     tds = tr.findall("db:td", namespaces=dbns)
  91     texts = [" ".join(x.replace('\u200b', '').replace('\u00b5', 'u').strip() for x in td.itertext() if x.strip() != '') for td in tds]
  92     return texts
  93
  94 data_rows = [get_texts_in_row(x) for x in data_trs]
  95 retired_cmd_rows = [get_texts_in_row(x) for x in retired_cmd_trs]
  96 cmd_rows = ([get_texts_in_row(x) for x in cmd_trs] +
  97             [x + ["RET"] for x in retired_cmd_rows])
  98
  99 def parse_tag(tag):
 100     # To handle some old cases where "x" is included as part of the tag number
 101     tag = tag.replace("x", "0")
 102     return f"0x{tag[1:5]}{tag[6:10]}"
 103 def parse_ret(ret):
 104     if ret.startswith("RET"):
 105         return -1
 106     else:
 107         return 0
 108 def include_in_parent_bit(name):
 109     if name in include_in_parent:
 110         return -1
 111     else:
 112         return 0
 113 def text_for_row(row):
 114     return f'    {{ {parse_tag(row[0])}, "{row[1]}", "{row[3]}", "{row[4]}", {parse_ret(row[5])}, {include_in_parent_bit(row[1])}}},'
 115
 116 def text_for_rows(rows):
 117     return "\n".join(text_for_row(row) for row in rows)
 118
 119 vrs = {i+1: get_texts_in_row(x)[0].split(maxsplit=1) for i,x in enumerate(get_trs(part05, "table_6.2-1"))}
 120
 121
 122 # Table A-1. UID Values
 123 uid_trs = get_trs(part06, "table_A-1")
 124 uid_rows = [get_texts_in_row(x) for x in uid_trs]
 125
 126 wkfr_trs = get_trs(part06, "table_A-2")
 127 wkfr_rows = [get_texts_in_row(x) for x in wkfr_trs]
 128 uid_rows += [x[:3] + ['Well-known frame of reference'] + x[3:] for x in wkfr_rows]
 129
 130 def uid_define_name(uid):
 131     if uid[1] == "(Retired)":
 132         return f'"{uid[0]}"'
 133     uid_type = uid[3]
 134     uid_name = uid[1]
 135     uid_name = re.sub(":.*", "", uid[1])
 136     if uid_name.endswith(uid_type):
 137         uid_name = uid_name[:-len(uid_type)].strip()
 138     return f"DCM_UID_{definify(uid_type)}_{definify(uid_name)}"
 139
 140 import re
 141 def definify(s):
 142     return re.sub('[^A-Z0-9]+', '_', re.sub('  +', ' ', re.sub('[^-A-Z0-9 ]+', '', s.upper())))
 143
 144 uid_rows = sorted(uid_rows, key=lambda uid_row: [int(i) for i in uid_row[0].split(".")])
 145 packet_dcm_h = """/* packet-dcm.h
 146  * Definitions for DICOM dissection
 147  * Copyright 2003, Rich Coe <richcoe2@gmail.com>
 148  * Copyright 2008-2018, David Aggeler <david_aggeler@hispeed.ch>
 149  *
 150  * DICOM communication protocol: https://www.dicomstandard.org/current/
 151  *
 152  * Generated automatically by """ + os.path.basename(sys.argv[0]) + """ from the following sources:
 153  *
 154  * """ + part05.find("./db:subtitle", namespaces=dbns).text + """
 155  * """ + part06.find("./db:subtitle", namespaces=dbns).text + """
 156  * """ + part07.find("./db:subtitle", namespaces=dbns).text + """
 157  *
 158  * Wireshark - Network traffic analyzer
 159  * By Gerald Combs <gerald@wireshark.org>
 160  * Copyright 1998 Gerald Combs
 161  *
 162  * SPDX-License-Identifier: GPL-2.0-or-later
 163  */
 164
 165 #ifndef __PACKET_DCM_H__
 166 #define __PACKET_DCM_H__
 167
 168 #ifdef __cplusplus
 169 extern "C" {
 170 #endif /* __cplusplus */
 171
 172 """ + "\n".join(f"#define DCM_VR_{vr[0]} {i:2d}  /* {vr[1]:25s} */" for i,vr in vrs.items()) + """
 173
 174 /* Following must be in the same order as the definitions above */
 175 static const char* dcm_tag_vr_lookup[] = {
 176     "  ",
 177     """ + ",\n    ".join(",".join(f'"{x[1][0]}"' for x in j[1]) for j in itertools.groupby(vrs.items(), lambda i: (i[0]-1)//8)) + """
 178 };
 179
 180
 181 /* ---------------------------------------------------------------------
 182  * DICOM Tag Definitions
 183  *
 184  * Some Tags can have different VRs
 185  *
 186  * Group 1000 is not supported, multiple tags with same description  (retired anyhow)
 187  * Group 7Fxx is not supported, multiple tags with same description  (retired anyhow)
 188  *
 189  * Tags (0020,3100 to 0020, 31FF) not supported, multiple tags with same description  (retired anyhow)
 190  *
 191  * Repeating groups (50xx & 60xx) are manually added. Declared as 5000 & 6000
 192  */
 193
 194 typedef struct dcm_tag {
 195     const uint32_t tag;
 196     const char *description;
 197     const char *vr;
 198     const char *vm;
 199     const bool is_retired;
 200     const bool add_to_summary;          /* Add to parent's item description */
 201 } dcm_tag_t;
 202
 203 static dcm_tag_t const dcm_tag_data[] = {
 204
 205     /* Command Tags */
 206 """ + text_for_rows(cmd_rows) + """
 207
 208     /* Data Tags */
 209 """ + text_for_rows(data_rows) + """
 210 };
 211
 212 /* ---------------------------------------------------------------------
 213  * DICOM UID Definitions
 214
 215  * Part 6 lists following different UID Types (2006-2008)
 216
 217  * Application Context Name
 218  * Coding Scheme
 219  * DICOM UIDs as a Coding Scheme
 220  * LDAP OID
 221  * Meta SOP Class
 222  * SOP Class
 223  * Service Class
 224  * Transfer Syntax
 225  * Well-known Print Queue SOP Instance
 226  * Well-known Printer SOP Instance
 227  * Well-known SOP Instance
 228  * Well-known frame of reference
 229  */
 230
 231 typedef struct dcm_uid {
 232     const char *value;
 233     const char *name;
 234     const char *type;
 235 } dcm_uid_t;
 236
 237 """ + "\n".join(f'#define {uid_define_name(uid)} "{uid[0]}"'
 238                 for uid in uid_rows if uid[1] != '(Retired)') + """
 239
 240 static dcm_uid_t const dcm_uid_data[] = {
 241 """ + "\n".join(f'    {{ {uid_define_name(uid)}, "{uid[1]}", "{uid[3]}"}},'
 242                             for uid in uid_rows)+ """
 243 };
 244
 245 #ifdef __cplusplus
 246 }
 247 #endif /* __cplusplus */
 248
 249 #endif /* packet-dcm.h */"""
 250
 251 print(packet_dcm_h)