Adarsh-Me commented on code in PR #3192: URL: https://github.com/apache/hugegraph/pull/3192#discussion_r4073487299
########## hugegraph-server/hugegraph-dist/docker/props.awk: ########## @@ -0,0 +1,280 @@ +# Licensed to the Apache Software Foundation (ASF) under one or more +# contributor license agreements. See the NOTICE file distributed with +# this work for additional information regarding copyright ownership. +# The ASF licenses this file to You under the Apache License, Version 2.0 +# (the "License"); you may not use this file except in compliance with +# the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. +# +# props.awk — read and rewrite Java ".properties" files with the grammar +# HugeConfig (commons-configuration over JDK Properties) applies, so the +# entrypoint and the server agree on what a mounted file means. grep/sed +# rewrites do not: they see `\`-escaped keys, `:` separators, continuation +# lines and duplicate definitions differently, which is how a mounted +# config ends up with two definitions of one key. +# +# One invocation, selected with the `PROPS_MODE` environment variable: +# +# PROPS_MODE=get PROPS_KEY=K PROPS_FILE=F +# print the value of K's first logical definition +# PROPS_MODE=set PROPS_KEY=K PROPS_FILE=F +# replace K's first definition in place, drop every other +# definition of K, append one when the file has none. The new +# value arrives pre-encoded in PROPS_VALUE_ENCODED (an environment +# variable, so secrets never appear in `ps` output or in awk's +# argv), and -v is not used for it so awk cannot mangle its +# backslash escapes. +# +# Grammar implemented (java.util.Properties line reader + the +# first-definition-wins rule Configuration.getString applies): +# - '#' / '!' comments and blank lines +# - '=' / ':' / whitespace separators, with whitespace then an optional +# single '=' or ':' accepted as one separator +# - continuations: a physical line ending in an odd number of +# backslashes joins the next line (its leading whitespace stripped) +# - backslash escapes in keys and values, including \uXXXX +# - duplicate logical keys resolve to the first definition +# +# Rewrites keep every untouched line byte-for-byte (comments, blank +# lines, unrelated entries), and replace the first definition where it +# stands, so mounted configs stay reviewable in git diffs. + +function die(msg) { + printf "props.awk: %s\n", msg > "/dev/stderr" + exit 1 +} + +function hex_digit(c) { + return index("0123456789abcdef", tolower(c)) - 1 +} + +# \uXXXX is a UTF-16 code unit in Java. Values here are effectively +# ISO-8859-1, so codes above 0xFF are kept as their literal escape text +# rather than being mangled through a single-byte sprintf. +function unescape(s, out, i, n, c, code, j, d, ok) { + out = "" + n = length(s) + for (i = 1; i <= n; i++) { + c = substr(s, i, 1) + if (c != "\\") { out = out c; continue } + if (i == n) break + i++ + c = substr(s, i, 1) + if (c == "u" && i + 4 <= n) { + code = 0 + ok = 1 + for (j = 1; j <= 4; j++) { + d = hex_digit(substr(s, i + j, 1)) + if (d < 0) { ok = 0; break } + code = code * 16 + d + } + if (ok) { + i += 4 + if (code <= 255) out = out sprintf("%c", code) + else out = out substr(s, i - 5, 6) + continue + } + } + if (c == "t") out = out "\t" + else if (c == "n") out = out "\n" + else if (c == "r") out = out "\r" + else if (c == "f") out = out "\f" + else out = out c + } + return out +} + +# A physical line is continued when it ends in an odd number of +# backslashes (an even count escapes itself). +function trailing_backslashes(s, n, k) { + n = length(s) + k = 0 + while (k < n && substr(s, n - k, 1) == "\\") k++ + return k +} + +function is_skipped(raw) { + return raw ~ /^[ \t]*([#!]|$)/ +} + +# Split a logical line into its raw (still-escaped) key and value parts. +# Results land in K_RAW / V_RAW because awk returns one value. +function split_kv(s, n, i, c, esc, sep_at, rest) { + n = length(s) + esc = 0 + sep_at = 0 + for (i = 1; i <= n; i++) { + c = substr(s, i, 1) + if (esc) { esc = 0; continue } + if (c == "\\") { esc = 1; continue } + if (c == "=" || c == ":" || c == " " || c == "\t") { sep_at = i; break } + } + if (sep_at == 0) { + K_RAW = s + V_RAW = "" + return + } + K_RAW = substr(s, 1, sep_at - 1) + rest = substr(s, sep_at) + c = substr(rest, 1, 1) + if (c == "=" || c == ":") { + rest = substr(rest, 2) + } else { + sub(/^[ \t]+/, "", rest) + c = substr(rest, 1, 1) + if (c == "=" || c == ":") rest = substr(rest, 2) + } + sub(/^[ \t]+/, "", rest) + V_RAW = rest +} + +function shquote(s) { + gsub(/'/, "'\\''", s) + return "'" s "'" +} + +# Load `file` into per-block arrays: one block per comment/blank line or +# logical entry, spanning exactly the physical lines it occupies. +function props_load(file, raw, rc, nl, stripped, next_raw, start, logical) { + NLINES = 0 + while ((rc = (getline raw < file)) > 0) { + NLINES++ + RAW[NLINES] = raw + } + if (rc == -1) + die("cannot read " file) + close(file) + + NBLOCK = 0 + for (nl = 1; nl <= NLINES; nl++) { + raw = RAW[nl] + # CRLF: java.util.Properties drops the line terminator, so one + # trailing CR is stripped for parsing only. RAW[] keeps the byte + # so props_set replays untouched lines byte-for-byte. + stripped = raw + sub(/\r$/, "", stripped) + if (is_skipped(stripped)) { + NBLOCK++ + BTYPE[NBLOCK] = "skip" + BFIRST[NBLOCK] = nl + BLAST[NBLOCK] = nl + continue + } + start = nl + logical = stripped + while (trailing_backslashes(logical) % 2 == 1 && nl < NLINES) { + logical = substr(logical, 1, length(logical) - 1) + nl++ + next_raw = RAW[nl] + sub(/\r$/, "", next_raw) + sub(/^[ \t]+/, "", next_raw) + logical = logical next_raw + } + # java.util.Properties ignores whitespace before the key; strip it + # so split_kv's separator scan agrees (an indented key used to be + # read as a key whose name started with a space, and a set then + # appended a second definition of the real key). + sub(/^[ \t]+/, "", logical) + split_kv(logical) + NBLOCK++ + BTYPE[NBLOCK] = "entry" + BFIRST[NBLOCK] = start + BLAST[NBLOCK] = nl + BKEY[NBLOCK] = unescape(K_RAW) + # Values stay in their on-disk escaped form. get Prop callers feed + # the result straight back into set, which would corrupt a decoded + # value by re-writing its backslashes as literals; keys are + # unescaped because they are matched against plain names. + BVAL[NBLOCK] = V_RAW + } +} + +function props_set(file, key, enc_val, tmp, cmd, b, first, ln) { + props_load(file) + first = 0 + for (b = 1; b <= NBLOCK; b++) { + if (BTYPE[b] == "entry" && BKEY[b] == key) { + if (first == 0) first = b + else BDROP[b] = 1 + } + } + # Staged rewrite: everything lands in a sibling temp file first, so a + # failure before the copy-back leaves the original untouched. The temp + # file can hold secrets, so it is pre-created 0600 before the first + # write: awk's `>` below would otherwise create it under the process + # umask (usually 0644), leaving auth.admin_pa or auth.token_secret + # briefly group- and world-readable. Truncating an existing file keeps + # its mode, and the chmod after close repairs a stale tmp left behind + # by a crashed run. + tmp = file ".tmp" + system("umask 077 && : > " shquote(tmp)) + for (b = 1; b <= NBLOCK; b++) { + if (BDROP[b]) continue + if (b == first) { + printf "%s=%s\n", key, enc_val > tmp + } else { + for (ln = BFIRST[b]; ln <= BLAST[b]; ln++) + print RAW[ln] > tmp + } + } + if (first == 0) + printf "%s=%s\n", key, enc_val > tmp + close(tmp) + # Copy the completed temp file back onto the original instead of + # renaming it: a rename replaces the inode, which would lose the + # file's permissions (a 0600 config holding secrets would come back + # umask-world-readable), turn a symlinked config into a regular file, + # and fail with EBUSY on a config bind-mounted as a single file — the + # mounted case this path exists for. The copy keeps the inode, mode, + # symlink and mount point, and since the temp file is fully written + # before the original is truncated, a failed copy still leaves the + # previous content on disk. + system("chmod 600 -- " shquote(tmp)) + cmd = "cat -- " shquote(tmp) " > " shquote(file) " && rm -f -- " shquote(tmp) Review Comment: Fixed in `b93b52e`. You are right, and the comment above it was the part that was wrong: `cat -- tmp > file` truncates the destination in the shell before `cat` runs, so an error mid-copy left a partial file and the "previous content is still on disk" claim did not hold. The copy-back is now preceded by a snapshot and followed by a repair: ```awk bak = file ".bak" cmd = "umask 077 && cp -- " shquote(file) " " shquote(bak) if (system(cmd " && chmod 600 -- " shquote(bak)) != 0) die("cannot back up " file " before the copy-back") cmd = "cat -- " shquote(tmp) " > " shquote(file) if (system(cmd) != 0) { msg = "cannot copy " tmp " over " file cmd = "cat -- " shquote(bak) " > " shquote(file) if (system(cmd) == 0) die(msg "; the previous content is restored") die(msg "; " file " is damaged, previous content is in " bak) } if (system("rm -f -- " shquote(tmp) " " shquote(bak)) != 0) die("cannot remove " tmp " and " bak " after the copy-back") ``` Notes on the choices: * Still a copy, not a rename, so the inode / mode / symlink / bind-mount properties the previous commit added stay true. * `umask 077` plus the explicit `chmod 600` means a backup taken of a 0644 mounted config is never *more* permissive than its source, and repairs a stale `.bak` from a crashed run the same way the tmp path already does. The snapshot can hold `auth.admin_pa` in clear. * Both staging files survive a failure on purpose: the `.tmp` is what was being written and the `.bak` is the way back. The success path removes both, which is now asserted. Evidence — `awk -f props.awk` driven by the entrypoint's `set_prop`, with the staging `cat` replaced through `PATH` so the copy fails the way `ENOSPC` would (stdout *is* the already-truncated destination, so the fake writes 23 bytes and exits 1): ``` fake cat invoked: -- .../config-rollback.tmp -- .../config-rollback.bak RESULT: set_prop failed as intended --- config after rollback --- auth.authenticator=org.apache.hugegraph.auth.StandardAuthenticator auth.token_secret=s3cr3t unrelated=true MATCH: content restored ``` Two cases carry this; `bf83032` adds the second one: the restore path above, and the case where the restore also fails — then the config is left damaged (0 bytes, measured; the honest limit of a copy that cannot be made atomic without replacing the inode), props.awk names `<file>.bak` in its error, and the snapshot is asserted byte-identical to the original. Reverting only `props.awk` to `bf2718f` makes the first case fail with `a failed copy-back must leave the previous content in place`, so the test does discriminate. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
