Drop a filter that matched nothing, keep the 79 ports it hid, and exit on a value the registers should not hold

data-import/httpmethod.py

@@ -3,13 +3,14 @@
 
     data-import/httpmethod.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
 
-A method whose name is not letters and hyphens does not ship, which drops the `*` the
-registry holds to stop anyone registering it.
+A row needs a name of letters and hyphens, which drops the registry's `*`, and both of
+the answers the registry gives for it.
 """
 import argparse
 import csv
 import io
 import re
+import sys
 from pathlib import Path
 
 import source
@@ -19,13 +20,19 @@ SOURCE = "https://www.iana.org/assignments/http-methods/methods.csv"
 OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "httpmethod.tsv"
 CACHE = Path(__file__).resolve().parent / "cache"
 COLUMNS = ["idempotent", "method", "safe"]
+ANSWERS = ("yes", "no")
 
 
 def rows(text):
     for r in csv.DictReader(io.StringIO(text)):
-        method, safe, idempotent = r["Method Name"].strip(), r["Safe"].strip(), r["Idempotent"].strip()
-        if re.match(r"^[A-Za-z][A-Za-z-]*$", method) and safe and idempotent:
-            yield {"idempotent": idempotent, "method": method, "safe": safe}
+        method = (r["Method Name"] or "").strip()
+        safe, idempotent = (r["Safe"] or "").strip(), (r["Idempotent"] or "").strip()
+        if not re.fullmatch(r"[A-Za-z][A-Za-z-]*", method) or not safe or not idempotent:
+            continue
+        for column, answer in (("Safe", safe), ("Idempotent", idempotent)):
+            if answer not in ANSWERS:
+                sys.exit(f"{method}: {column} {answer!r} is neither yes nor no")
+        yield {"idempotent": idempotent, "method": method, "safe": safe}
 
 
 def main():

data-import/port.py

@@ -3,13 +3,14 @@
 
     data-import/port.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
 
-What ships is the TCP assignments in use: a row needs a service name, a port number and
-a description the registry has filled in. A port the registry lists more than once keeps
-the first service, so a number selects one row.
+A row needs a service name and a numeric TCP port. A port the registry lists more than
+once keeps the first service it describes, or the first of them where it describes none,
+so a number selects one row.
 """
 import argparse
 import csv
 import io
+import sys
 from pathlib import Path
 
 import source
@@ -18,21 +19,24 @@ import tsv
 SOURCE = "https://www.iana.org/assignments/service-names-port-numbers/service-names-port-numbers.csv"
 OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "port.tsv"
 CACHE = Path(__file__).resolve().parent / "cache"
-COLUMNS = ["name", "number"]
-UNUSED = ("Reserved", "Unassigned", "IANA assigned this well-formed service name to replace an unregistered or squatted port.")
+COLUMNS = ["number", "service"]
+FLOOR = 4000
 
 
 def rows(text):
-    seen = set()
+    best, described = {}, set()
     for r in csv.DictReader(io.StringIO(text)):
-        name, number = r["Service Name"].strip(), r["Port Number"].strip()
-        described = (r["Description"] or "").strip()
-        if r["Transport Protocol"].strip() != "tcp" or not name or not number.isdigit():
+        service, number = (r["Service Name"] or "").strip(), (r["Port Number"] or "").strip()
+        transport, description = (r["Transport Protocol"] or "").strip(), (r["Description"] or "").strip()
+        if transport != "tcp" or not service or not number.isdigit():
             continue
-        if not described or described in UNUSED or number in seen:
-            continue
-        seen.add(number)
-        yield {"name": name, "number": number}
+        if not 1 <= int(number) <= 65535:
+            sys.exit(f"{service}: port {number} is outside 1-65535")
+        if number not in best or (description and number not in described):
+            best[number] = {"number": number, "service": service}
+        if description:
+            described.add(number)
+    return sorted(best.values(), key=lambda r: int(r["number"]))
 
 
 def main():
@@ -41,7 +45,9 @@ def main():
     p.add_argument("--source", default=SOURCE)
     p.add_argument("--out", default=str(OUT))
     a = p.parse_args()
-    table = sorted(rows(source.fetch(a.source, a.cache, "service-names-port-numbers.csv").decode("utf-8")), key=lambda r: int(r["number"]))
+    table = rows(source.fetch(a.source, a.cache, "service-names-port-numbers.csv").decode("utf-8"))
+    if len(table) < FLOOR:
+        sys.exit(f"only {len(table)} ports named a TCP service; the registry's columns have moved")
     tsv.write(a.out, COLUMNS, table)
 
 

data-import/protocol.py

@@ -3,12 +3,14 @@
 
     data-import/protocol.py [--source URL_OR_FILE] [--cache DIR] [--out FILE]
 
-A number the registry leaves unassigned, reserves or marks deprecated does not ship.
-Where it names no protocol beside the keyword, the keyword is the name.
+A row needs a number and a keyword, which drops the seven numbers assigned to a class of
+protocols rather than to one; a keyword the registry reserves or marks deprecated goes
+too. Where the registry names no protocol beside the keyword, the keyword is the name.
 """
 import argparse
 import csv
 import io
+import sys
 from pathlib import Path
 
 import source
@@ -18,15 +20,28 @@ SOURCE = "https://www.iana.org/assignments/protocol-numbers/protocol-numbers-1.c
 OUT = Path(__file__).resolve().parent.parent / "data" / "misc" / "protocol.tsv"
 CACHE = Path(__file__).resolve().parent / "cache"
 COLUMNS = ["keyword", "name", "number"]
-SKIP = ("Unassigned", "Reserved", "deprecated")
+SKIP = ("Reserved", "deprecated")
 
 
 def rows(text):
     for r in csv.DictReader(io.StringIO(text)):
-        number, keyword = r["Decimal"].strip(), r["Keyword"].strip()
+        number, keyword = (r["Decimal"] or "").strip(), (r["Keyword"] or "").strip()
         if not number.isdigit() or not keyword or any(s in keyword for s in SKIP):
             continue
-        yield {"keyword": keyword, "name": " ".join(r["Protocol"].split()) or keyword, "number": number}
+        if not 0 <= int(number) <= 255:
+            sys.exit(f"{keyword}: protocol number {number} is outside 0-255")
+        yield {"keyword": keyword, "name": " ".join((r["Protocol"] or "").split()) or keyword, "number": number}
+
+
+def refuse_a_name_the_loader_would(table):
+    """A name spelling another row's keyword, or a second row's name, is a load error for every consumer."""
+    keywords, seen = {r["keyword"] for r in table}, {}
+    for r in table:
+        if r["name"] != r["keyword"] and r["name"] in keywords:
+            sys.exit(f"{r['keyword']}: name {r['name']!r} is another protocol's keyword")
+        if r["name"] in seen:
+            sys.exit(f"{r['keyword']}: name {r['name']!r} repeats {seen[r['name']]}'s")
+        seen[r["name"]] = r["keyword"]
 
 
 def main():
@@ -36,6 +51,7 @@ def main():
     p.add_argument("--out", default=str(OUT))
     a = p.parse_args()
     table = sorted(rows(source.fetch(a.source, a.cache, "protocol-numbers-1.csv").decode("utf-8")), key=lambda r: int(r["number"]))
+    refuse_a_name_the_loader_would(table)
     tsv.write(a.out, COLUMNS, table)