{"entries":[{"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-07-27T15:03:49.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-07-27T15:03:49.000000+01:00"},"hash":"292e7203b4a196dc91a3570af55348fae25e92a9","message":"fix(progress): keep the spinner status on a single row (#1500)\n\n* fix(progress): keep the spinner status on a single row\n\nEach spinner frame clears its line with `\\r\\x1b[K`, which only erases the\nrow the cursor sits on. When the status is wider than the terminal it\nwraps, so every redraw leaves the overflow row on screen and the line\nreads as flicker. The combined status\n`Refreshing model pricing from LiteLLM... :: Loading usage logs (0/2) ::\nClaude, Codex` is ~85 columns, so any terminal narrower than that hits it\non every online run; `-O` skips the pricing status and avoids it.\n\nTruncate the status to the terminal width minus the spinner prefix and one\nspare column before writing, so the single-line clear stays correct and\nthe cursor never ends a frame in the wrap-pending state.\n\n`table.rs` already had ANSI- and CJK-aware truncation for cells, so that\nhelper moves to `ccusage-terminal::width` as the public\n`truncate_to_width` instead of being duplicated. Its tests and insta\nsnapshot move with it.\n\nVerified under a pty at `COLUMNS=40`: all 284 spinner frames stay within\n39 columns, and at `COLUMNS=80` nothing is truncated.\n\nCloses #1411\n\n* fix(terminal): keep a zero-width truncation empty\n\n`truncate_to_width` treated a width of 0 like a width of 1 and returned the\nellipsis, which occupies one display column and so overran the budget it was\ngiven. The spinner reaches this on terminals of three columns or fewer,\nwhere `fit_status_to_width` has no room left after the spinner prefix, so\nthe status could still spill into the column that is deliberately kept\nspare.\n\nReturn an empty string instead, and fold the CSI-scanning loop that\n`visible_width` and `truncate_to_width` each carried into a shared\n`skip_ansi_escape` helper so the two cannot drift apart.\n\n---------\n\nCo-authored-by: Cursor Agent \nCo-authored-by: ryoppippi ","tree":"f385e68e7db2150228e40f0946e5ddb8a5fa4981"},"mode":"0100644","name":"lib.rs","size":242},{"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-07-27T15:03:49.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-07-27T15:03:49.000000+01:00"},"hash":"292e7203b4a196dc91a3570af55348fae25e92a9","message":"fix(progress): keep the spinner status on a single row (#1500)\n\n* fix(progress): keep the spinner status on a single row\n\nEach spinner frame clears its line with `\\r\\x1b[K`, which only erases the\nrow the cursor sits on. When the status is wider than the terminal it\nwraps, so every redraw leaves the overflow row on screen and the line\nreads as flicker. The combined status\n`Refreshing model pricing from LiteLLM... :: Loading usage logs (0/2) ::\nClaude, Codex` is ~85 columns, so any terminal narrower than that hits it\non every online run; `-O` skips the pricing status and avoids it.\n\nTruncate the status to the terminal width minus the spinner prefix and one\nspare column before writing, so the single-line clear stays correct and\nthe cursor never ends a frame in the wrap-pending state.\n\n`table.rs` already had ANSI- and CJK-aware truncation for cells, so that\nhelper moves to `ccusage-terminal::width` as the public\n`truncate_to_width` instead of being duplicated. Its tests and insta\nsnapshot move with it.\n\nVerified under a pty at `COLUMNS=40`: all 284 spinner frames stay within\n39 columns, and at `COLUMNS=80` nothing is truncated.\n\nCloses #1411\n\n* fix(terminal): keep a zero-width truncation empty\n\n`truncate_to_width` treated a width of 0 like a width of 1 and returned the\nellipsis, which occupies one display column and so overran the budget it was\ngiven. The spinner reaches this on terminals of three columns or fewer,\nwhere `fit_status_to_width` has no room left after the spinner prefix, so\nthe status could still spill into the column that is deliberately kept\nspare.\n\nReturn an empty string instead, and fold the CSI-scanning loop that\n`visible_width` and `truncate_to_width` each carried into a shared\n`skip_ansi_escape` helper so the two cannot drift apart.\n\n---------\n\nCo-authored-by: Cursor Agent \nCo-authored-by: ryoppippi ","tree":"f385e68e7db2150228e40f0946e5ddb8a5fa4981"},"mode":"0040000","name":"snapshots","size":0},{"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-05-23T14:55:29.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-05-23T14:55:29.000000+01:00"},"hash":"fe3f4805b677417e55a830e59e714ae7bdfb2d6a","message":"refactor(terminal): split terminal helpers into modules (#1127)\n\nBreak the ccusage-terminal crate out of the single lib.rs file while keeping the public exports unchanged.\n\nMove table rendering, terminal styling, title rendering, width handling, and terminal width discovery into focused private modules so future terminal work has clearer ownership.\n\nThe snapshot outputs are unchanged; their files move only because the tests now live beside the extracted modules.","tree":"a669547d47aacb92255661e419b56538fb06831c"},"mode":"0100644","name":"style.rs","size":1302},{"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-07-27T15:03:49.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-07-27T15:03:49.000000+01:00"},"hash":"292e7203b4a196dc91a3570af55348fae25e92a9","message":"fix(progress): keep the spinner status on a single row (#1500)\n\n* fix(progress): keep the spinner status on a single row\n\nEach spinner frame clears its line with `\\r\\x1b[K`, which only erases the\nrow the cursor sits on. When the status is wider than the terminal it\nwraps, so every redraw leaves the overflow row on screen and the line\nreads as flicker. The combined status\n`Refreshing model pricing from LiteLLM... :: Loading usage logs (0/2) ::\nClaude, Codex` is ~85 columns, so any terminal narrower than that hits it\non every online run; `-O` skips the pricing status and avoids it.\n\nTruncate the status to the terminal width minus the spinner prefix and one\nspare column before writing, so the single-line clear stays correct and\nthe cursor never ends a frame in the wrap-pending state.\n\n`table.rs` already had ANSI- and CJK-aware truncation for cells, so that\nhelper moves to `ccusage-terminal::width` as the public\n`truncate_to_width` instead of being duplicated. Its tests and insta\nsnapshot move with it.\n\nVerified under a pty at `COLUMNS=40`: all 284 spinner frames stay within\n39 columns, and at `COLUMNS=80` nothing is truncated.\n\nCloses #1411\n\n* fix(terminal): keep a zero-width truncation empty\n\n`truncate_to_width` treated a width of 0 like a width of 1 and returned the\nellipsis, which occupies one display column and so overran the budget it was\ngiven. The spinner reaches this on terminals of three columns or fewer,\nwhere `fit_status_to_width` has no room left after the spinner prefix, so\nthe status could still spill into the column that is deliberately kept\nspare.\n\nReturn an empty string instead, and fold the CSI-scanning loop that\n`visible_width` and `truncate_to_width` each carried into a shared\n`skip_ansi_escape` helper so the two cannot drift apart.\n\n---------\n\nCo-authored-by: Cursor Agent \nCo-authored-by: ryoppippi ","tree":"f385e68e7db2150228e40f0946e5ddb8a5fa4981"},"mode":"0100644","name":"table.rs","size":15018},{"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-05-23T18:17:24.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-05-23T18:17:24.000000+01:00"},"hash":"309a29524f4fc3f20c988fea632cb869b53abf1b","message":"fix(terminal): use terminal_size for width detection (#1134)\n\n* fix(terminal): use terminal_size for width detection\n\nReplace the hand-written Unix ioctl path with the terminal_size crate so terminal width detection can rely on maintained cross-platform code. This keeps the existing COLUMNS override and default width behaviour while allowing the crate to handle stdout, stderr, stdin, and Windows terminal handles.\n\nAdd focused tests for the width selection rules so invalid COLUMNS values, zero widths, and fallback behaviour remain explicit.\n\n* ci: avoid PR GitHub token for static Nix builds\n\nThe pull_request Linux arm static build repeatedly failed while fetching the public crane flake input through Nix with an HTTP 401 Bad credentials response. The same build path passed on the push workflow, so the failure is tied to the pull_request token that setup-nix configures for GitHub fetches.\n\nFor pull_request static Linux builds, clear Nix access-tokens for the build command so public GitHub flake inputs are fetched without the invalid token. Push builds keep the existing authenticated configuration.","tree":"aec76388adffed55b17c2d9abba6ef623c7ce23d"},"mode":"0100644","name":"terminal.rs","size":1235},{"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-06-13T14:52:53.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-06-13T06:52:53.000000-07:00"},"hash":"5816e640e8824986f33427203da164b01394a161","message":"build(rust): migrate crates to edition 2024 (#1308)\n\n* build(rust): migrate crates to edition 2024\n\nUpdate all Rust crate manifests from edition 2021 to 2024 now that the workspace toolchain is on Rust 1.96. Align treefmt rustfmt configuration with the crate edition so local formatting and flake checks use the same parser mode. Apply cargo fix, clippy --fix, and rustfmt output for edition-compatibility changes such as env var unsafe calls, expr_2021 macro fragments, let-chain collapses, and import ordering.\n\n* test(rust): centralize environment variable guards\n\nAdd a shared test-support environment variable guard that holds a single process-wide mutex while temporarily setting environment variables. Replace module-local locks and direct unsafe env mutations in tests with the shared RAII guard so Rust 2024 unsafe environment access is isolated in one audited helper and restored consistently.","tree":"0e2106340208b61bad154ba70424545dfd5b622b"},"mode":"0100644","name":"title.rs","size":2633},{"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-07-27T15:03:49.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-07-27T15:03:49.000000+01:00"},"hash":"292e7203b4a196dc91a3570af55348fae25e92a9","message":"fix(progress): keep the spinner status on a single row (#1500)\n\n* fix(progress): keep the spinner status on a single row\n\nEach spinner frame clears its line with `\\r\\x1b[K`, which only erases the\nrow the cursor sits on. When the status is wider than the terminal it\nwraps, so every redraw leaves the overflow row on screen and the line\nreads as flicker. The combined status\n`Refreshing model pricing from LiteLLM... :: Loading usage logs (0/2) ::\nClaude, Codex` is ~85 columns, so any terminal narrower than that hits it\non every online run; `-O` skips the pricing status and avoids it.\n\nTruncate the status to the terminal width minus the spinner prefix and one\nspare column before writing, so the single-line clear stays correct and\nthe cursor never ends a frame in the wrap-pending state.\n\n`table.rs` already had ANSI- and CJK-aware truncation for cells, so that\nhelper moves to `ccusage-terminal::width` as the public\n`truncate_to_width` instead of being duplicated. Its tests and insta\nsnapshot move with it.\n\nVerified under a pty at `COLUMNS=40`: all 284 spinner frames stay within\n39 columns, and at `COLUMNS=80` nothing is truncated.\n\nCloses #1411\n\n* fix(terminal): keep a zero-width truncation empty\n\n`truncate_to_width` treated a width of 0 like a width of 1 and returned the\nellipsis, which occupies one display column and so overran the budget it was\ngiven. The spinner reaches this on terminals of three columns or fewer,\nwhere `fit_status_to_width` has no room left after the spinner prefix, so\nthe status could still spill into the column that is deliberately kept\nspare.\n\nReturn an empty string instead, and fold the CSI-scanning loop that\n`visible_width` and `truncate_to_width` each carried into a shared\n`skip_ansi_escape` helper so the two cannot drift apart.\n\n---------\n\nCo-authored-by: Cursor Agent \nCo-authored-by: ryoppippi ","tree":"f385e68e7db2150228e40f0946e5ddb8a5fa4981"},"mode":"0100644","name":"width.rs","size":4980}],"lastCommit":{"author":{"email":"1560508+ryoppippi@users.noreply.github.com","name":"ryoppippi","when":"2026-07-27T15:03:49.000000+01:00"},"committer":{"email":"noreply@github.com","name":"GitHub","when":"2026-07-27T15:03:49.000000+01:00"},"hash":"292e7203b4a196dc91a3570af55348fae25e92a9","message":"fix(progress): keep the spinner status on a single row (#1500)\n\n* fix(progress): keep the spinner status on a single row\n\nEach spinner frame clears its line with `\\r\\x1b[K`, which only erases the\nrow the cursor sits on. When the status is wider than the terminal it\nwraps, so every redraw leaves the overflow row on screen and the line\nreads as flicker. The combined status\n`Refreshing model pricing from LiteLLM... :: Loading usage logs (0/2) ::\nClaude, Codex` is ~85 columns, so any terminal narrower than that hits it\non every online run; `-O` skips the pricing status and avoids it.\n\nTruncate the status to the terminal width minus the spinner prefix and one\nspare column before writing, so the single-line clear stays correct and\nthe cursor never ends a frame in the wrap-pending state.\n\n`table.rs` already had ANSI- and CJK-aware truncation for cells, so that\nhelper moves to `ccusage-terminal::width` as the public\n`truncate_to_width` instead of being duplicated. Its tests and insta\nsnapshot move with it.\n\nVerified under a pty at `COLUMNS=40`: all 284 spinner frames stay within\n39 columns, and at `COLUMNS=80` nothing is truncated.\n\nCloses #1411\n\n* fix(terminal): keep a zero-width truncation empty\n\n`truncate_to_width` treated a width of 0 like a width of 1 and returned the\nellipsis, which occupies one display column and so overran the budget it was\ngiven. The spinner reaches this on terminals of three columns or fewer,\nwhere `fit_status_to_width` has no room left after the spinner prefix, so\nthe status could still spill into the column that is deliberately kept\nspare.\n\nReturn an empty string instead, and fold the CSI-scanning loop that\n`visible_width` and `truncate_to_width` each carried into a shared\n`skip_ansi_escape` helper so the two cannot drift apart.\n\n---------\n\nCo-authored-by: Cursor Agent \nCo-authored-by: ryoppippi ","tree":"f385e68e7db2150228e40f0946e5ddb8a5fa4981"}}