feat: address open GitHub issues and parsedate PR

.github/workflows/ci.yml

@@ -59,6 +59,7 @@ jobs:
 
       - name: Run benchmarks
         id: bench
+        continue-on-error: true
         run: |
           BENCH_ARGS="--ci --markdown-file benchmark-report.md --json-file benchmark-results.json"
           if [ -f perf-baseline.json ]; then
@@ -77,11 +78,9 @@ jobs:
               console.log('No benchmark report found, skipping comment.');
               return;
             }
-            let body = fs.readFileSync(reportPath, 'utf-8').trim();
+            const body = fs.readFileSync(reportPath, 'utf-8').trim();
             if (!body) return;
 
-            body += '\n\n---\n*Compared against master baseline. Threshold: 25% regression.*';
-
             const { data: comments } = await github.rest.issues.listComments({
               owner: context.repo.owner,
               repo: context.repo.repo,

man/man1/recs-decollate.1

@@ -15,6 +15,9 @@ Reverse of collate: takes a single record and produces multiple records using de
 \fB--deaggregator\fR, \fB-d\fR \fI<deaggregators>\fR
 Deaggregator specification (colon\-separated).
 .TP
+\fB--only\fR, \fB-o\fR
+Only output deaggregated fields, excluding original record fields. Useful when you only want the expanded data, not the source record.
+.TP
 \fB--list-deaggregators\fR
 List available deaggregators and exit.
 
@@ -27,6 +30,14 @@ Split the \'hosts\' field into individual \'host\' fields
 .fi
 .RE
 
+Decollate and only keep deaggregated fields
+.PP
+.RS 4
+.nf
+\fBrecs decollate --only -d \'unarray,items,,item\'\fR
+.fi
+.RE
+
 .SH SEE ALSO
 \fBrecs\-collate\fR(1)
 

man/man1/recs-fromxls.1

@@ -0,0 +1,60 @@
+.TH RECS\-FROMXLS 1 "2026-02-22" "recs 0.1.0" "RecordStream Manual"
+
+.SH NAME
+recs\-fromxls \- Parse Excel files (xls, xlsx, xlsb, xlsm) into records
+
+.SH SYNOPSIS
+.B recs fromxls [options] <files...>
+
+.SH DESCRIPTION
+Parse Excel files (xls, xlsx, xlsb, xlsm) into records. By default, reads the first sheet and uses the first row as header names.
+.PP
+
+.SH OPTIONS
+.TP
+\fB--key\fR, \fB-k\fR \fI<keys>\fR
+Comma separated list of field names. Overrides header detection.
+.TP
+\fB--field\fR, \fB-f\fR \fI<keys>\fR
+Comma separated list of field names. Overrides header detection.
+.TP
+\fB--no-header\fR, \fB-n\fR
+Do not treat the first row as a header. Fields will be named numerically (0, 1, 2, ...).
+.TP
+\fB--sheet\fR, \fB-s\fR \fI<name>\fR
+Specify a sheet name to read. Defaults to the first sheet.
+.TP
+\fB--all-sheets\fR
+Read all sheets in the workbook, adding a \'sheet\' field to each record.
+
+.SH EXAMPLES
+Read an Excel file using headers from the first row
+.PP
+.RS 4
+.nf
+\fBrecs fromxls data.xlsx\fR
+.fi
+.RE
+
+Read a specific sheet without headers
+.PP
+.RS 4
+.nf
+\fBrecs fromxls --sheet \'Sheet2\' --no-header -k name,value data.xlsx\fR
+.fi
+.RE
+
+Read all sheets
+.PP
+.RS 4
+.nf
+\fBrecs fromxls --all-sheets data.xlsx\fR
+.fi
+.RE
+
+.SH SEE ALSO
+\fBrecs\-fromcsv\fR(1)
+
+.SH AUTHOR
+Benjamin Bernard <perlhacker@benjaminbernard.com>
+

man/man1/recs-multiplex.1

@@ -21,6 +21,12 @@ Domain language key: name=expression where the expression evaluates as a valuati
 \fB--line-key\fR, \fB-L\fR \fI<key>\fR
 Use the value of this key as line input for the nested operation (rather than the entire record). Use with recs from* operations generally.
 .TP
+\fB--output-file-key\fR, \fB-o\fR \fI<key>\fR
+Write each group\'s output to a separate file, using the value of the given key as the filename.
+.TP
+\fB--output-file-eval\fR, \fB-O\fR \fI<expression>\fR
+Write each group\'s output to a separate file, with filename determined by the given expression. Supports {{key}} interpolation with group key values.
+.TP
 \fB--adjacent\fR, \fB-1\fR
 Only group together adjacent records. Avoids spooling records into memory.
 .TP
@@ -59,6 +65,14 @@ Separate out a stream of text by PID into separate invocations of an operation
 .fi
 .RE
 
+Write each group\'s CSV output to separate files by department
+.PP
+.RS 4
+.nf
+\fBrecs multiplex -k department -O \'output-{{department}}.csv\' -- recs tocsv\fR
+.fi
+.RE
+
 .SH SEE ALSO
 \fBrecs\-collate\fR(1), \fBrecs\-chain\fR(1)
 

man/man1/recs-parsedate.1

@@ -0,0 +1,66 @@
+.TH RECS\-PARSEDATE 1 "2026-02-22" "recs 0.1.0" "RecordStream Manual"
+
+.SH NAME
+recs\-parsedate \- Parse date/time strings from a field and output them in a normalized format
+
+.SH SYNOPSIS
+.B recs parsedate [options] [files...]
+
+.SH DESCRIPTION
+Parse date/time strings from a field and output them in a normalized format. Supports epoch seconds, ISO 8601, and custom strftime\-style format strings for both input and output.
+.PP
+
+.SH OPTIONS
+.TP
+\fB--key\fR, \fB-k\fR \fI<key>\fR
+Key field containing the date/time string to parse.
+.TP
+\fB--output\fR, \fB-o\fR \fI<key>\fR
+Output key for the parsed date (defaults to \'parsed_<key>\').
+.TP
+\fB--format\fR, \fB-f\fR \fI<format>\fR
+Input format for parsing (strftime\-style). Common directives: %Y (4\-digit year), %m (month), %d (day), %H (hour 24h), %M (minute), %S (second), %b (abbreviated month name), %p (AM/PM).
+.TP
+\fB--output-format\fR, \fB-F\fR \fI<format>\fR
+Output format (strftime\-style). Defaults to ISO 8601 if not specified.
+.TP
+\fB--epoch\fR, \fB-e\fR
+Input date is in epoch seconds.
+.TP
+\fB--output-epoch\fR, \fB-E\fR
+Output as epoch seconds instead of a formatted string.
+.TP
+\fB--timezone\fR, \fB-z\fR \fI<timezone>\fR
+Timezone for output formatting (IANA name like \'America/New_York\').
+
+.SH EXAMPLES
+Parse dates and output as epoch seconds
+.PP
+.RS 4
+.nf
+\fBrecs parsedate -k date -E\fR
+.fi
+.RE
+
+Parse a custom date format and reformat it
+.PP
+.RS 4
+.nf
+\fBrecs parsedate -k timestamp -f \'%d/%b/%Y:%H:%M:%S\' -F \'%Y-%m-%d %H:%M:%S\'\fR
+.fi
+.RE
+
+Convert epoch seconds to ISO 8601
+.PP
+.RS 4
+.nf
+\fBrecs parsedate -k time -e -o iso_time\fR
+.fi
+.RE
+
+.SH SEE ALSO
+\fBrecs\-normalizetime\fR(1)
+
+.SH AUTHOR
+Benjamin Bernard <perlhacker@benjaminbernard.com>
+

man/man1/recs.1

@@ -53,6 +53,9 @@ Runs tcpdump and puts out records, one for each packet
 \fBrecs fromxferlog\fR
 Each line of input (or lines of <files>) is parsed as an FTP transfer log (xferlog format) to produce an output record
 .TP
+\fBrecs fromxls\fR
+Parse Excel files (xls, xlsx, xlsb, xlsm) into records
+.TP
 \fBrecs fromxml\fR
 Reads either from STDIN or from the specified URIs
 
@@ -100,6 +103,9 @@ Take records, grouped together by \-\-keys, and run a separate operation instanc
 \fBrecs normalizetime\fR
 Given a single key field containing a date/time value, construct a normalized version of the value and place it into a field named \'n_<key>\'
 .TP
+\fBrecs parsedate\fR
+Parse date/time strings from a field and output them in a normalized format
+.TP
 \fBrecs sort\fR
 Sort records from input or from files
 .TP

package.json

@@ -32,6 +32,8 @@
     "fast-xml-parser": "^5.3.7",
     "mongodb": "^7.1.0",
     "openai": "^6.22.0",
-    "papaparse": "^5.5.3"
+    "papaparse": "^5.5.3",
+    "string-width": "^8.2.0",
+    "xlsx": "^0.18.5"
   }
 }

src/Operation.ts

@@ -50,12 +50,18 @@ export class PrinterReceiver implements RecordReceiver {
  */
 export class CollectorReceiver implements RecordReceiver {
   readonly records: Record[] = [];
+  readonly lines: string[] = [];
 
   acceptRecord(record: Record): boolean {
     this.records.push(record);
     return true;
   }
 
+  acceptLine(line: string): boolean {
+    this.lines.push(line);
+    return true;
+  }
+
   finish(): void {
     // nothing to do
   }

src/cli/dispatcher.ts

@@ -20,6 +20,7 @@ import { FromRe } from "../operations/input/fromre.ts";
 import { FromSplit } from "../operations/input/fromsplit.ts";
 import { FromTcpdump } from "../operations/input/fromtcpdump.ts";
 import { FromXferlog } from "../operations/input/fromxferlog.ts";
+import { FromXls } from "../operations/input/fromxls.ts";
 import { FromXml } from "../operations/input/fromxml.ts";
 
 // -- Transform operations --
@@ -40,6 +41,7 @@ import { JoinOperation } from "../operations/transform/join.ts";
 import { CollateOperation } from "../operations/transform/collate.ts";
 import { DecollateOperation } from "../operations/transform/decollate.ts";
 import { ExpandJsonOperation } from "../operations/transform/expandjson.ts";
+import { ParseDateOperation } from "../operations/transform/parsedate.ts";
 import { ChainOperation, registerOperationFactory } from "../operations/transform/chain.ts";
 import { MultiplexOperation } from "../operations/transform/multiplex.ts";
 
@@ -75,6 +77,7 @@ const operationRegistry = new Map<string, OpConstructor>([
   ["fromsplit", FromSplit],
   ["fromtcpdump", FromTcpdump],
   ["fromxferlog", FromXferlog],
+  ["fromxls", FromXls],
   ["fromxml", FromXml],
   // Transform
   ["grep", GrepOperation],
@@ -94,6 +97,7 @@ const operationRegistry = new Map<string, OpConstructor>([
   ["join", JoinOperation],
   ["collate", CollateOperation],
   ["decollate", DecollateOperation],
+  ["parsedate", ParseDateOperation],
   ["chain", ChainOperation],
   ["multiplex", MultiplexOperation],
   // Output